Texture analysis of non-small cell lung cancer on unenhanced computed tomography: initial evidence for a relationship with tumour glucose metabolism and stage

Integrating intratumoral and peritumoral radiomics with deep transfer learning for DCE-MRI breast lesion differentiation: A multicenter study comparing performance with radiologists

PURPOSE

To conduct the fusion of radiomics and deep transfer learning features from the intratumoral and peritumoral areas in breast DCE-MRI images to differentiate between benign and malignant breast tumors, and to compare the diagnostic accuracy of this fusion model against the assessments made by experienced radiologists.

MATERIALS AND METHODS

This multi-center study conducted a retrospective analysis of DCE-MRI images from 330 women diagnosed with breast cancer, with 138 cases categorized as benign and 192 as malignant. The training and internal testing sets comprised 270 patients from center 1, while the external testing cohort consisted of 60 patients from center 2. A fusion feature set consisting of radiomics features and deep transfer learning features was constructed from both intratumoral (ITR) and peritumoral (PTR) areas. The Least absolute shrinkage and selection operator (LASSO) based support vector machine was chosen as the classifier by comparing its performance with five other machine learning models. The diagnostic performance and clinical usefulness of fusion model were verified and assessed through the area under the receiver operating characteristics (ROC) and decision curve analysis. Additionally, the performance of the fusion model was compared with the diagnostic assessments of two experienced radiologists to evaluate its relative accuracy. The study strictly adhered to CLEAR and METRICS guidelines for standardization to ensure rigorous and reproducible methods.

RESULTS

The findings show that the fusion model, utilizing radiomics and deep transfer learning features from the ITR and PTR, exhibited exceptional performance in classifying breast tumors, achieving AUCs of 0.950 in the internal testing set and 0.921 in the external testing set. This performance significantly surpasses that of models relying on singular regional radiomics or deep transfer learning features alone. Moreover, the fusion model demonstrated superior diagnostic accuracy compared to the evaluations conducted by two experienced radiologists, thereby highlighting its potential to support and enhance clinical decision-making in the differentiation of benign and malignant breast tumors.

CONCLUSION

The fusion model, combining multi-regional radiomics with deep transfer learning features, not only accurately differentiates between benign and malignant breast tumors but also outperforms the diagnostic assessments made by experienced radiologists. This underscores the model's potential as a valuable tool for improving the accuracy and reliability of breast tumor diagnosis.

CT-based radiomics and clinical characteristics for predicting bone metastasis in lung adenocarcinoma patients

Background

The occurrence of bone metastasis (BM) will seriously shorten the survival time of lung adenocarcinoma patients and aggravate the suffering of patients. Computed tomography (CT)-based clinical radiomics nomogram may help clinicians stratify the risk of BM in lung adenocarcinoma patients, thereby enabling personalized individualized clinical decision making.

Methods

A total of 501 patients with lung adenocarcinoma from March 2017 to March 2019 were enrolled in the study. Based on plain chest CT images, 1130 radiomics features were extracted from each lesion. One-way analysis of variance (ANOVA) and least absolute shrinkage selection operator (LASSO) algorithm were used for radiomics features selection. Univariate and multivariate analyses were used to screen for clinical characteristics and identify independent predictors of BM. Three models (radiomics model, clinical model and combined model) were constructed to predict BM in lung adenocarcinoma patients. Receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were used to evaluate the performance of the three models. The DeLong test was used to compare the performance of the models.

Results

Finally, the clinical model for predicting BM in lung adenocarcinoma patients was constructed based on 5 independent predictors: cytokeratin 19-fragments (CYFRA21-1), stage, Ki-67, edge, and lobulation. The radiomics model was constructed based on 5 radiomics features. The combined model incorporating clinical independent predictors and radiomics was constructed. In the validation cohort, the area under the curve (AUC) of the clinical model, radiomics model and combined model was 0.824, 0.842 and 0.866, respectively. Delong test showed that in the training cohort, the AUC values of the radiomics model and the combined model were statistically different (P=0.03), and the AUC values of the other models were not statistically different. DCA showed that the nomogram had a highest net clinical benefit.

Conclusions

The CT-based clinical radiomics nomogram can be used as a non-invasive and quantitative method to help clinicians stratify the risk of BM in patients with lung adenocarcinoma, thereby enabling personalized clinical decision making.

Prediction of transformation in the histopathological growth pattern of colorectal liver metastases after chemotherapy using CT-based radiomics

Chemotherapy alters the prognostic biomarker histopathological growth pattern (HGP) phenotype in colorectal liver metastases (CRLMs) patients. We aimed to develop a CT-based radiomics model to predict the transformation of the HGP phenotype after chemotherapy. This study included 181 patients with 298 CRLMs who underwent preoperative contrast-enhanced CT followed by partial hepatectomy between January 2007 and July 2022 at two institutions. HGPs were categorized as pure desmoplastic HGP (pdHGP) or non-pdHGP. The samples were allocated to training, internal validation, and external validation cohorts comprising 153, 65, and 29 CRLMs, respectively. Radiomics analysis was performed on pre-enhanced, arterial phase, portal venous phase (PVP), and fused images. The model was used to predict prechemotherapy HGPs in 112 CRLMs, and HGP transformation was analysed by comparing these findings with postchemotherapy HGPs determined pathologically. The prevalence of pdHGP was 19.8% (23/116) and 45.8% (70/153) in chemonaïve and postchemotherapy patients, respectively (P < 0.001). The PVP radiomics signature showed good performance in distinguishing pdHGP from non-pdHGPs (AUCs of 0.906, 0.877, and 0.805 in the training, internal validation, and external validation cohorts, respectively). The prevalence of prechemotherapy pdHGP predicted by the radiomics model was 33.0% (37/112), and the prevalence of postchemotherapy pdHGP according to the pathological analysis was 47.3% (53/112; P = 0.029). The transformation of HGP was bidirectional, with 15.2% (17/112) of CRLMs transforming from prechemotherapy pdHGP to postchemotherapy non-pdHGP and 30.4% (34/112) transforming from prechemotherapy non-pdHGP to postchemotherapy pdHGP (P = 0.005). CT-based radiomics method can be used to effectively predict the HGP transformation in chemotherapy-treated CRLM patients, thereby providing a basis for treatment decisions.

Novel model integrating computed tomography-based image markers with genetic markers for discriminating radiation pneumonitis in patients with unresectable stage III non-small cell lung cancer receiving radiotherapy: a retrospective multi-center radiogenomics study

BACKGROUND

Chemoradiotherapy is a critical treatment for patients with locally advanced and unresectable non-small cell lung cancer (NSCLC), and it is essential to identify high-risk patients as early as possible owing to the high incidence of radiation pneumonitis (RP). Increasing attention is being paid to the effects of endogenous factors for RP. This study aimed to investigate the value of computed tomography (CT)-based radiomics combined with genomics in analyzing the risk of grade ≥ 2 RP in unresectable stage III NSCLC.

METHODS

In this retrospective multi-center observational study, 100 patients with unresectable stage III NSCLC who were treated with chemoradiotherapy were analyzed. Radiomics features of the entire lung were extracted from pre-radiotherapy CT images. The least absolute shrinkage and selection operator algorithm was used for optimal feature selection to calculate the Rad-score for predicting grade ≥ 2 RP. Genomic DNA was extracted from formalin-fixed paraffin-embedded pretreatment biopsy tissues. Univariate and multivariate logistic regression analyses were performed to identify predictors of RP for model development. The area under the receiver operating characteristic curve was used to evaluate the predictive capacity of the model. Statistical comparisons of the area under the curve values between different models were performed using the DeLong test. Calibration and decision curves were used to demonstrate discriminatory and clinical benefit ratios, respectively.

RESULTS

The Rad-score was constructed from nine radiomic features to predict grade ≥ 2 RP. Multivariate analysis demonstrated that histology, Rad-score, and XRCC1 (rs25487) allele mutation were independent high-risk factors correlated with RP. The area under the curve of the integrated model combining clinical factors, radiomics, and genomics was significantly higher than that of any single model (0.827 versus 0.594, 0.738, or 0.641). Calibration and decision curve analyses confirmed the satisfactory clinical feasibility and utility of the nomogram.

CONCLUSION

Histology, Rad-score, and XRCC1 (rs25487) allele mutation could predict grade ≥ 2 RP in patients with locally advanced unresectable NSCLC after chemoradiotherapy, and the integrated model combining clinical factors, radiomics, and genomics demonstrated the best predictive efficacy.

A user-friendly deep learning application for accurate lung cancer diagnosis

BACKGROUND

Accurate diagnosis and subsequent delineated treatment planning require the experience of clinicians in the handling of their case numbers. However, applying deep learning in image processing is useful in creating tools that promise faster high-quality diagnoses, but the accuracy and precision of 3-D image processing from 2-D data may be limited by factors such as superposition of organs, distortion and magnification, and detection of new pathologies. The purpose of this research is to use radiomics and deep learning to develop a tool for lung cancer diagnosis.

METHODS

This study applies radiomics and deep learning in the diagnosis of lung cancer to help clinicians accurately analyze the images and thereby provide the appropriate treatment planning. 86 patients were recruited from Bach Mai Hospital, and 1012 patients were collected from an open-source database. First, deep learning has been applied in the process of segmentation by U-NET and cancer classification via the use of the DenseNet model. Second, the radiomics were applied for measuring and calculating diameter, surface area, and volume. Finally, the hardware also was designed by connecting between Arduino Nano and MFRC522 module for reading data from the tag. In addition, the displayed interface was created on a web platform using Python through Streamlit.

RESULTS

The applied segmentation model yielded a validation loss of 0.498, a train loss of 0.27, a cancer classification validation loss of 0.78, and a training accuracy of 0.98. The outcomes of the diagnostic capabilities of lung cancer (recognition and classification of lung cancer from chest CT scans) were quite successful.

CONCLUSIONS

The model provided means for storing and updating patients' data directly on the interface which allowed the results to be readily available for the health care providers. The developed system will improve clinical communication and information exchange. Moreover, it can manage efforts by generating correlated and coherent summaries of cancer diagnoses.

18 F-FDG PET/MRI of Primary Hepatic Malignancies: Differential Diagnosis and Histologic Grading

BACKGROUND

Distinguishing between IHCC and HCC is important because of their differences in treatment and prognosis. The hybrid Positron Emission Tomography/magnetic Resonance Imaging (PET/MRI) system has become more widely accessible, with oncological imaging becoming one of its most promising applications.

OBJECTIVE

The objective of this study was to see how well 18F-fluorodeoxyglucose (18F-FDG) PET/MRI could be used for differential diagnosis and histologic grading of primary hepatic malignancies.

METHODS

We retrospectively evaluated 64 patients (53 patients with HCC, 11 patients with IHCC) with histologically proven primary hepatic malignancies using 18F-FDG/MRI. The Apparent Diffusion Coefficient (ADC), Coefficient of Variance (CV) of the ADC, and standardized uptake value (SUV) were calculated.

RESULTS

The mean SUVmax value was higher for IHCC (7.7 ± 3.4) than for HCC (5.2 ± 3.1) (p = 0.019). The area under the curve (AUC) was 0.737, an optimal 6.98 cut-off value providing 72% sensitivity and 79% specificity. The ADCcv value in IHCC was statistically significantly higher than in HCC (p=0.014). ADC mean values in HCCs were significantly higher in low-grade tumors than in high-grade tumors. The AUC value was 0.73, and the optimal cut-off point was 1.20x10-6 mm2/s, giving 62% sensitivity and 72% specificity. The SUVmax value was also found to be statistically significantly higher in the high-grade group. The ADCcv value in the HCC low-grade group was found to be lower than in the highgrade group (p=0.036).

CONCLUSION

18F FDG PET/MRI is a novel imaging technique that can aid in the differentiation of primary hepatic neoplasms as well as tumor-grade estimation.

CT Texture Analysis in Nonalcoholic Fatty Liver Disease (NAFLD)

Background

Nonalcoholic fatty liver disease (NAFLD) is the most common form of liver disease worldwide. There are limited biomarkers that can detect progression from simple steatosis to nonalcoholic steatohepatitis (NASH). The purpose of our study was to utilize CT texture analysis to distinguish steatosis from NASH.

Methods

16 patients with NAFLD (38% male, median (interquartile range): age 57 (48-64) years, BMI 37.5 (35.0-46.8) kg/m2) underwent liver biopsy and abdominal non-contrast CT. CT texture analysis was performed to quantify gray-level tissue summaries (e.g., entropy, kurtosis, skewness, and attenuation) using commercially available software (TexRad, Cambridge England). Logistic regression analyses were performed to quantify the association between steatosis/NASH status and CT texture. ROC curve analysis was performed to determine sensitivity, specificity, AUC, 95% CIs, and cutoff values of texture parameters to differentiate steatosis from NASH.

Results

By histology, 6/16 (37%) of patients had simple steatosis and 10/16 (63%) had NASH. Patients with NASH had lower entropy (median, interquartile range (IQR): 4.3 (4.1, 4.8) vs. 5.0 (4.9, 5.2), P = 0.013) and lower mean value of positive pixels (MPP) (34.4 (21.8, 52.2) vs. 66.5 (57.0, 70.7), P = 0.009) than those with simple steatosis. Entropy values below 4.73 predict NASH with 100% (95%CI: 67-100%) specificity and 80% (50-100%) sensitivity, AUC: 0.88. MPP values below 54.0 predict NASH with 100% (67-100%) specificity and 100% (50-100%) sensitivity, AUC 0.90.

Conclusion

Our study provides preliminary evidence that CT texture analysis may serve as a novel imaging biomarker for disease activity in NAFLD and the discrimination of steatosis and NASH.

MRI-based Radiomics Model for Preoperative Prediction of Lateral Pelvic Lymph Node Metastasis in Locally Advanced Rectal Cancer

RATIONALE AND OBJECTIVES

To develop a magnetic resonance imaging (MRI)-based radiomics model for preoperative prediction of lateral pelvic lymph node (LPLN) metastasis (LPLNM) in patients with locally advanced rectal cancer MATERIALS AND METHODS: We retrospectively enrolled 263 patients with rectal cancer who underwent total mesorectal excision and LPLN dissection. Radiomics features from the primary lesion and LPLNs on baseline MRI images were utilized to construct a radiomics model, and their radiomics scores were combined to develop a radiomics scoring system. A clinical prediction model was developed using logistic regression. A hybrid predicting model was created through multivariable logistic regression analysis, integrating the radiomics score with significant clinical risk factors (baseline Carcinoembryonic Antigen (CEA), clinical circumferential resection margin status, and the short axis diameter of LPLN). This hybrid model was presented with a hybrid clinical-radiomics nomogram, and its calibration, discrimination, and clinical usefulness were assessed.

RESULTS

A total of 148 patients were included in the analysis and randomly divided into a training cohort (n = 104) and an independent internal testing cohort (n = 44). The hybrid clinical-radiomics model exhibited the highest discrimination, with an area under the receiver operating characteristic (AUC) of 0.843 [95% confidence interval (CI), 0.706-0.968] in the testing cohort compared to the clinical model [AUC (95% CI) = 0.772 (0.589-0.856)] and radiomics model [AUC (95% CI) = 0.731 (0.613-0.849)]. The hybrid prediction model also demonstrated good calibration, and decision curve analysis confirmed its clinical usefulness.

CONCLUSION

This study developed a hybrid MRI-based radiomics model that incorporates a combination of radiomics score and significant clinical risk factors. The proposed model holds promise for individualized preoperative prediction of LPLNM in patients with locally advanced rectal cancer.

DATA AVAILABILITY STATEMENT

The data presented in this study are available on request from the corresponding author.

Three dimensional computed tomography texture analysis of pulmonary lesions: Does radiomics allow differentiation between carcinoma, neuroendocrine tumor and organizing pneumonia?

PURPOSE

To investigate whether CT texture analysis allows differentiation between adenocarcinomas, squamous cell carcinomas, carcinoids, small cell lung cancers and organizing pneumonia and between carcinomas and neuroendocrine tumors.

METHOD

This retrospective study included patients 133 patients (30 patients with organizing pneumonia, 30 patients with adenocarcinoma, 30 patients with squamous cell carcinoma, 23 patients with small cell lung cancer, 20 patients with carcinoid), who underwent CT-guided biopsy of the lung and had a corresponding histopathologic diagnosis. Pulmonary lesions were segmented in consensus by two radiologists with and without a threshold of -50HU in three dimensions. Groupwise comparisons were performed to assess for differences between all five above-listed entities and between carcinomas and neuroendocrine tumors.

RESULTS

Pairwise comparisons of the five entities revealed 53 statistically significant texture features when using no HU-threshold and 6 statistically significant features with a threshold of -50HU. The largest AUC (0.818 [95%CI 0.706-0.930]) was found for the feature wavelet-HHH_glszm_SmallAreaEmphasis for discrimination of carcinoid from the other entities when using no HU-threshold. In differentiating neuroendocrine tumors from carcinomas, 173 parameters proved statistically significant when using no HU threshold versus 52 parameters when using a -50HU-threshold. The largest AUC (0.810 [95%CI 0.728-0,893]) was found for the parameter original_glcm_Correlation for discrimination of neuroendocrine tumors from carcinomas when using no HU-threshold.

CONCLUSIONS

CT texture analysis revealed features that differed significantly between malignant pulmonary lesions and organizing pneumonia and between carcinomas and neuroendocrine tumors of the lung. Applying a HU-threshold for segmentation substantially influenced the results of texture analysis.

Radiomics in colorectal cancer patients

The main therapeutic options for colorectal cancer are surgical resection and adjuvant chemotherapy in non-metastatic disease. However, the evaluation of the overall adjuvant chemotherapy benefit in patients with a high risk of recurrence is challenging. Radiological images can represent a source of data that can be analyzed by using automated computer-based techniques, working on numerical information coded within Digital Imaging and Communications in Medicine files: This image numerical analysis has been named "radiomics". Radiomics allows the extraction of quantitative features from radiological images, mainly invisible to the naked eye, that can be further analyzed by artificial intelligence algorithms. Radiomics is expanding in oncology to either understand tumor biology or for the development of imaging biomarkers for diagnosis, staging, and prognosis, prediction of treatment response and diseases monitoring and surveillance. Several efforts have been made to develop radiomics signatures for colorectal cancer patient using computed tomography (CT) images with different aims: The preoperative prediction of lymph node metastasis, detecting BRAF and RAS gene mutations. Moreover, the use of delta-radiomics allows the analysis of variations of the radiomics parameters extracted from CT scans performed at different timepoints. Most published studies concerning radiomics and magnetic resonance imaging (MRI) mainly focused on the response of advanced tumors that underwent neoadjuvant therapy. Nodes status is the main determinant of adjuvant chemotherapy. Therefore, several radiomics model based on MRI, especially on T2-weighted images and ADC maps, for the preoperative prediction of nodes metastasis in rectal cancer has been developed. Current studies mostly focused on the applications of radiomics in positron emission tomography/CT for the prediction of survival after curative surgical resection and assessment of response following neoadjuvant chemoradiotherapy. Since colorectal liver metastases develop in about 25% of patients with colorectal carcinoma, the main diagnostic tasks of radiomics should be the detection of synchronous and metachronous lesions. Radiomics could be an additional tool in clinical setting, especially in identifying patients with high-risk disease. Nevertheless, radiomics has numerous shortcomings that make daily use extremely difficult. Further studies are needed to assess performance of radiomics in stratifying patients with high-risk disease.

CT texture analysis reliability in pulmonary lesions: the influence of 3D vs. 2D lesion segmentation and volume definition by a Hounsfield-unit threshold

OBJECTIVE

Reproducibility problems are a known limitation of radiomics. The segmentation of the target lesion plays a critical role in texture analysis variability. This study's aim was to compare the interobserver reliability of manual 2D vs. 3D lung lesion segmentation with and without pre-definition of the volume using a threshold of - 50 HU.

METHODS

Seventy-five patients with histopathologically proven lung lesions (15 patients each with adenocarcinoma, squamous cell carcinoma, small cell lung cancer, carcinoid, and organizing pneumonia) who underwent an unenhanced CT scan of the chest were included. Three radiologists independently segmented each lesion manually in 3D and 2D with and without pre-segmentation volume definition by a HU threshold, and shape parameters and original, Laplacian of Gaussian-filtered, and wavelet-based texture features were derived. To assess interobserver reliability and identify the most robust texture features, intraclass correlation coefficients (ICCs) for different segmentation settings were calculated.

RESULTS

Shape parameters had high reliability (64-79% had excellent and good ICCs). Texture features had weak reliability levels, with the highest ICCs (38% excellent or good) found for original features in 3D segmentation without the use of a HU threshold. A small proportion (4.3-11.5%) of texture features had excellent or good ICC values at all segmentation settings.

CONCLUSION

Interobserver reliability of texture features from CT scans of a heterogeneous collection of manually segmented lung lesions was low with a small proportion of features demonstrating high reliability independent of the segmentation settings. These results indicate a limited applicability of texture analysis and the need to define robust texture features in patients with lung lesions.

KEY POINTS

• Our study showed a low reproducibility of texture features when 3 radiologists independently segmented lung lesions in CT images, which highlights a serious limitation of texture analysis. • Interobserver reliability of texture features was low regardless of whether the lesion was segmented in 2D and 3D with or without a HU threshold. • In contrast to texture features, shape parameters showed a high interobserver reliability when lesions were segmented in 2D vs. 3D with and without a HU threshold of - 50.

Radiomics approach with deep learning for predicting T4 obstructive colorectal cancer using CT image

OBJECTIVES

Patients with T4 obstructive colorectal cancer (OCC) have a high mortality rate. Therefore, an accurate distinction between T4 and T1-T3 (NT4) in OCC is an important part of preoperative evaluation, especially in the emergency setting. This paper introduces three models of radiomics, deep learning, and deep learning-based radiomics to identify T4 OCC.

METHODS

We established a dataset of computed tomography (CT) images of 164 patients with pathologically confirmed OCC, from which 2537 slides were extracted. First, since T4 tumors penetrate the bowel wall and involve adjacent organs, we explored whether the peritumoral region contributes to the assessment of T4 OCC. Furthermore, we visualized the radiomics and deep learning features using the t-distributed stochastic neighbor embedding technique (t-SNE). Finally, we built a merged model by fusing radiomic features with deep learning features. In this experiment, the performance of each model was evaluated by the area under the receiver operating characteristic curve (AUC).

RESULTS

In the test cohort, the AUC values predicted by the radiomics model in the dilated region of interest (dROI) was 0.770. And the AUC value of the deep learning model with the patches extended 20-pixel reached 0.936. Combining the characteristics of radiomics and deep learning, our method achieved an AUC value of 0.947 in the T4 and non-T4 (NT4) classification, and increased the AUC value to 0.950 after the addition of clinical features.

CONCLUSION

The prediction results of our merged model of deep learning radiomics outperformed the deep learning model and significantly outperformed the radiomics model. The experimental results demonstrate that combining the peritumoral region improves the prediction performance of the radiomics model and the deep learning model.

Prediction of Ki-67 of Invasive Ductal Breast Cancer Based on Ultrasound Radiomics Nomogram

PURPOSE

The objective of this research was to develop and validate an ultrasound-based radiomics nomogram for the pre-operative assessment of Ki-67 in breast cancer (BC).

MATERIALS AND METHODS

From December 2016 to December 2018, 515 patients with invasive ductal breast cancer who received two-dimensional (2D) ultrasound and Ki-67 examination were studied and analyzed retrospectively. The dataset was distributed at random into a training cohort (n = 360) and a test cohort (n = 155) in the ratio of 7:3. Each tumor region of interest was defined based on 2D ultrasound images and radiomics features were extracted. ANOVA, maximum correlation minimum redundancy (mRMR) algorithm, and minimum absolute shrinkage and selection operator (LASSO) were performed to pick features, and independent clinical predictors were integrated with radscore to construct the nomogram for predicting Ki-67 index by univariate and multivariate logistic regression analysis. The performance and utility of the models were evaluated by plotting receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and calibration curves.

RESULTS

In the testing cohort, the area under the receiver characteristic curve (AUC) of the nomogram was 0.770 (95% confidence interval, 0.690-0.860). In both cohorts, the nomogram outperformed both the clinical model and the radiomics model (P < .05 according to the DeLong test). The analysis of DCA proved that the model has clinical utility.

CONCLUSIONS

The nomogram based on 2D ultrasound images offered an approach for predicting Ki-67 in BC.

Quantitative characterization of duodenal gastrinoma autofluorescence using multiphoton microscopy

BACKGROUND

Duodenal gastrinomas (DGASTs) are neuroendocrine tumors that develop in the submucosa of the duodenum and produce the hormone gastrin. Surgical resection of DGASTs is complicated by the small size of these tumors and the tendency for them to develop diffusely in the duodenum. Endoscopic mucosal resection of DGASTs is an increasingly popular method for treating this disease due to its low complication rate but suffers from poor rates of pathologically negative margins. Multiphoton microscopy can capture high-resolution images of biological tissue with contrast generated from endogenous fluorescence (autofluorescence [AF]) through two-photon excited fluorescence (2PEF). Second harmonic generation is another popular method of generating image contrast with multiphoton microscopy (MPM) and is a light-scattering phenomenon that occurs predominantly from structures such as collagen in biological samples. Some molecules that contribute to AF change in abundance from processes related to the cancer disease process (e.g., metabolic changes, oxidative stress, and angiogenesis).

STUDY DESIGN/MATERIALS AND METHODS

MPM was used to image 12 separate patient samples of formalin-fixed and paraffin-embedded duodenal gastrinoma slides with a second-harmonic generation (SHG) channel and four 2PEF channels. The excitation and emission profiles of each 2PEF channel were tuned to capture signal dominated by distinct fluorophores with well-characterized fluorescent spectra and known connections to the physiologic changes that arise in cancerous tissue.

RESULTS

We found that there was a significant difference in the relative abundance of signal generated in the 2PEF channels for regions of DGASTs in comparison to the neighboring tissues of the duodenum. Data generated from texture feature extraction of the MPM images were used in linear discriminant analysis models to create classifiers for tumor versus all other tissue types before and after principal component analysis (PCA). PCA improved the classifier accuracy and reduced the number of features required to achieve maximum accuracy. The linear discriminant classifier after PCA distinguished between tumor and other tissue types with an accuracy of 90.6%-93.8%.

CONCLUSIONS

These results suggest that multiphoton microscopy 2PEF and SHG imaging is a promising label-free method for discriminating between DGASTs and normal duodenal tissue which has implications for future applications of in vivo assessment of resection margins with endoscopic MPM.

CT-based machine learning radiomics predicts CCR5 expression level and survival in ovarian cancer

OBJECTIVE

We aimed to evaluate the prognostic value of C-C motif chemokine receptor type 5 (CCR5) expression level for patients with ovarian cancer and to establish a radiomics model that can predict CCR5 expression level using The Cancer Imaging Archive (TCIA) and The Cancer Genome Atlas (TCGA) database.

METHODS

A total of 343 cases of ovarian cancer from the TCGA were used for the gene-based prognostic analysis. Fifty seven cases had preoperative computed tomography (CT) images stored in TCIA with genomic data in TCGA were used for radiomics feature extraction and model construction. 89 cases with both TCGA and TCIA clinical data were used for radiomics model evaluation. After feature extraction, a radiomics signature was constructed using the least absolute shrinkage and selection operator (LASSO) regression analysis. A prognostic scoring system incorporating radiomics signature based on CCR5 expression level and clinicopathologic risk factors was proposed for survival prediction.

RESULTS

CCR5 was identified as a differentially expressed prognosis-related gene in tumor and normal sample, which were involved in the regulation of immune response and tumor invasion and metastasis. Four optimal radiomics features were selected to predict overall survival. The performance of the radiomics model for predicting the CCR5 expression level with 10-fold cross- validation achieved Area Under Curve (AUCs) of 0.770 and of 0.726, respectively, in the training and validation sets. A predictive nomogram was generated based on the total risk score of each patient, the AUCs of the time-dependent receiver operating characteristic (ROC) curve of the model was 0.8, 0.673 and 0.792 for 1-year, 3-year and 5-year, respectively. Along with clinical features, important imaging biomarkers could improve the overall survival accuracy of the prediction model.

CONCLUSION

The expression levels of CCR5 can affect the prognosis of patients with ovarian cancer. CT-based radiomics could serve as a new tool for prognosis prediction.

Computer-aided diagnostic models to classify lymph node metastasis and lymphoma involvement in enlarged cervical lymph nodes using PET/CT

BACKGROUND

It is a clinical problem to identify histological component in enlarged cervical lymph nodes, particularly in differentiation between lymph node metastasis and lymphoma involvement.

PURPOSE

To construct two kinds of deep learning (DL)-based computer-aided diagnosis (CAD) systems including DL-convolutional neural networks (DL-CNN) and DL-machine learning for pathological diagnosis of cervical lymph nodes by positron emission tomography (PET)/computed tomography (CT) images.

METHODS

We collected CT, PET, and PET/CT images series from 165 patients with enlarged cervical lymph nodes receiving examinations from January 2014 to June 2018. Six CNNs pretrained on ImageNet as DL architectures were used for two kinds of DL-based CAD models, including DL-CNN and DL-machine learning models. The DL-CNN models were constructed via transfer learning for classification of lymphomatous and metastatic lymph nodes. The DL-machine learning models were developed by DL-based features extractors and support vector machine (SVM) classifier. As for DL-SVM models, we also evaluate the effect of handcrafted radiomics features in combination of DL-based features.

RESULTS

The DL-CNN model with ResNet50 architecture on PET/CT images had the best diagnostic performance among all six algorithms with an area under the receiver operating characteristic curve (AUC) of 0.845 and accuracy of 78.13% in the testing cohort. The DL-SVM model on ResNet50 extractor showed great performance for the testing cohort with an AUC of 0.901, accuracy of 86.96%, sensitivity of 76.09%, and specificity of 94.20%. The combination of DL-based and handcrafted features yielded the improvement of diagnostic performance.

CONCLUSIONS

Our DL-based CAD systems on PET/CT images were developed for classifying metastatic and lymphomatous involvement with favorable diagnostic performance in enlarged cervical lymph nodes. Further clinical practice of our systems may improve quality of the following therapeutic interventions and optimize patients' outcomes.

Fractal and textural imaging identify new subgroups of patients with colorectal cancer based on biophysical properties of the cancer cells

Colorectal cancer (CRC) can been sub-divided, based on the generation of tertiary lymphoid structures (TLS), into CRC with a Crohn's like lymphoid reaction (CLR) representing de novo formation of TLSs or CRC lacking TLSs that show Diffuse Inflammatory infiltration (DII). The association between TLS, early treatment initiation and longer survival highlights the need for deeper patient stratification that could lead to more targeted therapies. We hypothesized that such stratification might be achieved by using digital image analyses. Here we retrospectively analyzed 35 CRC patient samples classified as CLR or DII by digital analysis, focusing on the parameters Fractal dimension, Lacunarity and the textural features Angular second momentum, Correlation, Inverse difference momentum and Entropy. Significant differences in the grades of these parameters between the two patient groups provided preliminary data that additional biophysical information can divide CRC into at least 3 subgroups which encompass CLR and DII. Additional studies are needed to test if this sub-classification aids in the selection of targeted therapy for patients with CRC.

Quantifying Tumor Heterogeneity from Multiparametric Magnetic Resonance Imaging of Prostate Using Texture Analysis

Simple Summary

Prostate cancer (PCa) occurs in males at a rate of 21.8%, predominantly at the customary primary site. High cure rates are possible through early detection and therapy when the tumor is still restricted to the prostate. These tumors do not grow rapidly, allowing for periods of up to 20 years between diagnosis and death. Multiparametric MRI (mp-MRI) is used as a non-invasive approach to diagnose PCa in subjects. This imaging method uses MR imaging with at least one functional MRI sequence to detect and characterize PCa. The use of multiparametric magnetic resonance imaging has refined the diagnosis of prostate cancer in radiology. Malignancy-modified critical features in tissue composition, such as heterogeneity, are associated with adverse tumor biology. Heterogeneity can be quantified through texture analysis, an effective technique for reviewing tumor images acquired in routine clinical practice. This study focused on identifying and quantifying tumor heterogeneity from prostate mp-MRI utilizing texture analysis.

Abstract

(1) Background: Multiparametric MRI (mp-MRI) is used to manage patients with PCa. Tumor identification via irregular sampling or biopsy is problematic and does not allow the comprehensive detection of the phenotypic and genetic alterations in a tumor. A non-invasive technique to clinically assess tumor heterogeneity is also in demand. We aimed to identify tumor heterogeneity from multiparametric magnetic resonance images using texture analysis (TA). (2) Methods: Eighteen patients with prostate cancer underwent mp-MRI scans before prostatectomy. A single radiologist matched the histopathology report to single axial slices that best depicted tumor and non-tumor regions to generate regions of interest (ROIs). First-order statistics based on the histogram analysis, including skewness, kurtosis, and entropy, were used to quantify tumor heterogeneity. We compared non-tumor regions with significant tumors, employing the two-tailed Mann–Whitney U test. Analysis of the area under the receiver operating characteristic curve (ROC-AUC) was used to determine diagnostic accuracy. (3) Results: ADC skewness for a 6 × 6 px filter was significantly lower with an ROC-AUC of 0.82 (p = 0.001). The skewness of the ADC for a 9 × 9 px filter had the second-highest result, with an ROC-AUC of 0.66; however, this was not statistically significant (p = 0.08). Furthermore, there were no substantial distinctions between pixel filter size groups from the histogram analysis, including entropy and kurtosis. (4) Conclusions: For all filter sizes, there was poor performance in terms of entropy and kurtosis histogram analyses for cancer diagnosis. Significant prostate cancer may be distinguished using a textural feature derived from ADC skewness with a 6 × 6 px filter size.

Using Radiomics as Prior Knowledge for Thorax Disease Classification and Localization in Chest X-rays

Chest X-ray becomes one of the most common medical diagnoses due to its noninvasiveness. The number of chest X-ray images has skyrocketed, but reading chest X-rays still have been manually performed by radiologists, which creates huge burnouts and delays. Traditionally, radiomics, as a subfield of radiology that can extract a large number of quantitative features from medical images, demonstrates its potential to facilitate medical imaging diagnosis before the deep learning era. In this paper, we develop an end-to-end framework, ChexRadiNet, that can utilize the radiomics features to improve the abnormality classification performance. Specifically, ChexRadiNet first applies a light-weight but efficient triplet-attention mechanism to classify the chest X-rays and highlight the abnormal regions. Then it uses the generated class activation map to extract radiomic features, which further guides our model to learn more robust image features. After a number of iterations and with the help of radiomic features, our framework can converge to more accurate image regions. We evaluate the ChexRadiNet framework using three public datasets: NIH ChestX-ray, CheXpert, and MIMIC-CXR. We find that ChexRadiNet outperforms the state-of-the-art on both disease detection (0.843 in AUC) and localization (0.679 in T(IoU) = 0.1). We make the code publicly available at https://github. com/bionlplab/lung_disease_detection_amia2021, with the hope that this method can facilitate the development of automatic systems with a higher-level understanding of the radiological world.

Can the computed tomography texture analysis of colorectal liver metastases predict the response to first-line cytotoxic chemotherapy?

BACKGROUND

Artificial intelligence in radiology has the potential to assist with the diagnosis, prognostication and therapeutic response prediction of various cancers. A few studies have reported that texture analysis can be helpful in predicting the response to chemotherapy for colorectal liver metastases, however, the results have varied. Necrotic metastases were not clearly excluded in these studies and in most studies the full range of texture analysis features were not evaluated. This study was designed to determine if the computed tomography (CT) texture analysis results of non-necrotic colorectal liver metastases differ from previous reports. A larger range of texture features were also evaluated to identify potential new biomarkers.

AIM

To identify potential new imaging biomarkers with CT texture analysis which can predict the response to first-line cytotoxic chemotherapy in non-necrotic colorectal liver metastases (CRLMs).

METHODS

Patients who presented with CRLMs from 2012 to 2020 were retrospectively selected on the institutional radiology information system of our private radiology practice. The inclusion criteria were non-necrotic CRLMs with a minimum size of 10 mm (diagnosed on archived 1.25 mm portal venous phase CT scans) which were treated with standard first-line cytotoxic chemotherapy (FOLFOX, FOLFIRI, FOLFOXIRI, CAPE-OX, CAPE-IRI or capecitabine). The final study cohort consisted of 29 patients. The treatment response of the CRLMs was classified according to the RECIST 1.1 criteria. By means of CT texture analysis, various first and second order texture features were extracted from a single non-necrotic target CRLM in each responding and non-responding patient. Associations between features and response to chemotherapy were assessed by logistic regression models. The prognostic accuracy of selected features was evaluated by using the area under the curve.

RESULTS

There were 15 responders (partial response) and 14 non-responders (7 stable and 7 with progressive disease). The responders presented with a higher number of CRLMs (P = 0.05). In univariable analysis, eight texture features of the responding CRLMs were associated with treatment response, but due to strong correlations among some of the features, only two features, namely minimum histogram gradient intensity and long run low grey level emphasis, were included in the multiple analysis. The area under the receiver operating characteristic curve of the multiple model was 0.80 (95%CI: 0.64 to 0.96), with a sensitivity of 0.73 (95%CI: 0.48 to 0.89) and a specificity of 0.79 (95%CI: 0.52 to 0.92).

CONCLUSION

Eight first and second order texture features, but particularly minimum histogram gradient intensity and long run low grey level emphasis are significantly correlated with treatment response in non-necrotic CRLMs.

Preliminary exploration of response the course of radiotherapy for stage III non-small cell lung cancer based on longitudinal CT radiomics features

Purpose

Explore the longitudinal CT-based radiomics to demonstrate the changing trend of radiotherapy response and to determine at which point after the onset of treatment radiomics exhibit the greatest change for stage III NSCLC patients.

Methods and materials

Ten stage III NSCLC patients in line with inclusion criteria were enrolled retrospectively, each of whom received radiotherapy or concurrent chemo-radiotherapy and performed eight series of follow-up CT imaging. Longitudinal radiomics were extracted on region of interest from the eight registered images, then two steps were conducted to select significant features as indicators of tumor change: 1) stable features were selected by Kendall rank correlation; 2) texture feature types with a steadily changing trend were retained and intensity features with stable change trends were selected to represent the large number of them. Next, the trend and rate of tumor change were analyzed using the Delta method and Curve-fitting method. Finally, the statistics in the distribution of stable features in patients were calculated.

Results

675 stable features were selected from a total number of 1371 radiomics features, then 12 texture features types were retained and three intensity features were chosen to represent their own category. Among the final selected feature types, it was found that the two time points were weeks 1 and 3 with the higher rate of change. One patient had very few stable tumor features out of a total of 101 features, and the rate of change of features of another patient was conspicuously higher than the average level with number of 301 features.

Conclusion

The longitudinal CT radiomics could demonstrate the change trend of tumor and at which point exhibit the greatest change during radiotherapy, and potentially be used for treatment decisions concerning adaptive radiotherapy.

Prediction of the Invasiveness of Ground-Glass Nodules in Lung Adenocarcinoma by Radiomics Analysis Using High-Resolution Computed Tomography Imaging

Background

Pure ground-glass nodules (pGGNs) have been considered inert tumors due to their biological behavior; however, their prognosis is not completely consistent because of differences in internal pathological component. The aim of this study was to explore whether radiomics can be used to identify the invasiveness of pGGNs.

Methods

The retrospective study received the relevant ethical approval. After postoperative pathological confirmation, sixty-five patients with lung adenocarcinoma pGGNs (≤30 mm) were enrolled in this study from January 2015 to October 2018. All the cases were randomly divided into training and test groups in a 7:3 ratio. In total, 385 radiomics features were obtained from HRCT images, and then least absolute shrinkage and selection operator (LASSO) logistic regression was applied to the training group to obtain optimal features to distinguish the invasion degree of lesions. The diagnostic efficiency of the radiomics model was estimated by the area under the curve (AUC) of the receiver operating curve (ROC), and verified by the test group.

Results

The optimal features (“GLCMEntropy_angle135_offset1” and “Sphericity”) were selected after applying the LASSO regression to develop the proposed radiomics model. This prediction model exhibited good differentiation between pre-invasive and invasive lesions. The AUC for the test group was 0.824 (95%CI: 0.599-1.000), indicating that the radiomics model has some prediction ability.

Conclusion

The HRCT radiomics features can discriminate pre-invasive from invasive lung adenocarcinoma pGGNs. This non-invasive method can provide more information for surgeons before operation, and can also predict the prognosis of patients to some extent.

Non-contrast-enhanced CT texture analysis of primary and metastatic pancreatic ductal adenocarcinomas: value in assessment of histopathological grade and differences between primary and metastatic lesions

PURPOSE

To evaluate the utility of non-contrast-enhanced CT texture analysis (CTTA) for predicting the histopathological differentiation of pancreatic ductal adenocarcinomas (PDAC) and to compare non-contrast-enhanced CTTA texture features between primary PDAC and hepatic metastases of PDAC.

METHODS

This retrospective study included 120 patients with histopathologically confirmed PDAC. Sixty-five patients underwent CT-guided biopsy of primary PDAC, while 55 patients underwent CT-guided biopsy of hepatic PDAC metastasis. All lesions were segmented in non-contrast-enhanced CT scans for CTTA based on histogram analysis, co-occurrence matrix, and run-length matrix. Statistical analysis was conducted for 372 texture features using Mann-Whitney U test, Bonferroni-Holm correction, and receiver operating characteristic (ROC) analysis. A p value < 0.05 was considered statistically significant.

RESULTS

Three features were identified that differed significantly between histopathological G2 and G3 primary tumors. Of these, "low gray-level zone emphasis" yielded the largest AUC (0.87 ± 0.04), reaching a sensitivity and specificity of 0.76 and 0.83, respectively, when a cut-off value of 0.482 was applied. Fifty-four features differed significantly between primary and hepatic metastatic PDAC.

CONCLUSION

Non-contrast-enhanced CTTA of PDAC identified differences in texture features between primary G2 and G3 tumors that could be used for non-invasive tumor assessment. Extensive differences between the features of primary and metastatic PDAC on CTTA suggest differences in tumor microenvironment.

Interpretable machine learning for predicting pathologic complete response in patients treated with chemoradiation therapy for rectal adenocarcinoma

Purpose

Pathologic complete response (pCR) is a critical factor in determining whether patients with rectal cancer (RC) should have surgery after neoadjuvant chemoradiotherapy (nCRT). Currently, a pathologist's histological analysis of surgical specimens is necessary for a reliable assessment of pCR. Machine learning (ML) algorithms have the potential to be a non-invasive way for identifying appropriate candidates for non-operative therapy. However, these ML models' interpretability remains challenging. We propose using explainable boosting machine (EBM) to predict the pCR of RC patients following nCRT.

Methods

A total of 296 features were extracted, including clinical parameters (CPs), dose-volume histogram (DVH) parameters from gross tumor volume (GTV) and organs-at-risk, and radiomics (R) and dosiomics (D) features from GTV. R and D features were subcategorized into shape (S), first-order (L1), second-order (L2), and higher-order (L3) local texture features. Multi-view analysis was employed to determine the best set of input feature categories. Boruta was used to select all-relevant features for each input dataset. ML models were trained on 180 cases from our institution, with 37 cases from RTOG 0822 clinical trial serving as the independent dataset for model validation. The performance of EBM in predicting pCR on the test dataset was evaluated using ROC AUC and compared with that of three state-of-the-art black-box models: extreme gradient boosting (XGB), random forest (RF) and support vector machine (SVM). The predictions of all black-box models were interpreted using Shapley additive explanations.

Results

The best input feature categories were CP+DVH+S+R_L1+R_L2 for all models, from which Boruta-selected features enabled the EBM, XGB, RF, and SVM models to attain the AUCs of 0.820, 0.828, 0.828, and 0.774, respectively. Although EBM did not achieve the best performance, it provided the best capability for identifying critical turning points in response scores at distinct feature values, revealing that the bladder with maximum dose >50 Gy, and the tumor with maximum2DDiameterColumn >80 mm, elongation <0.55, leastAxisLength >50 mm and lower variance of CT intensities were associated with unfavorable outcomes.

Conclusions

EBM has the potential to enhance the physician's ability to evaluate an ML-based prediction of pCR and has implications for selecting patients for a "watchful waiting" strategy to RC therapy.

Impact of Interobserver Variability in Manual Segmentation of Non-Small Cell Lung Cancer (NSCLC) Applying Low-Rank Radiomic Representation on Computed Tomography

Simple Summary

Discovery of predictive and prognostic radiomic features in cancer is currently of great interest to the radiologic and oncologic community. Tumor phenotypic and prognostic information can be obtained by extracting features on tumor segmentations, and it is typically imaging analysts, physician trainees, and attending physicians who provide these labeled datasets for analysis. The potential impact of level and type of specialty training on interobserver variability in manual segmentation of NSCLC was examined. Although there was some variability in segmentation between readers, the subsequently extracted radiomic features were overall well correlated. High fidelity radiomic feature extraction relies on accurate feature extraction from imaging that produce robust prognostic and predictive radiomic NSCLC biomarkers. This study concludes that this goal can be obtained using segmenters of different levels of training and clinical experience.

Abstract

This study tackles interobserver variability with respect to specialty training in manual segmentation of non-small cell lung cancer (NSCLC). Four readers included for segmentation are: a data scientist (BY), a medical student (LS), a radiology trainee (MH), and a specialty-trained radiologist (SK) for a total of 293 patients from two publicly available databases. Sørensen–Dice (SD) coefficients and low rank Pearson correlation coefficients (CC) of 429 radiomics were calculated to assess interobserver variability. Cox proportional hazard (CPH) models and Kaplan-Meier (KM) curves of overall survival (OS) prediction for each dataset were also generated. SD and CC for segmentations demonstrated high similarities, yielding, SD: 0.79 and CC: 0.92 (BY-SK), SD: 0.81 and CC: 0.83 (LS-SK), and SD: 0.84 and CC: 0.91 (MH-SK) in average for both databases, respectively. OS through the maximal CPH model for the two datasets yielded c-statistics of 0.7 (95% CI) and 0.69 (95% CI), while adding radiomic and clinical variables (sex, stage/morphological status, and histology) together. KM curves also showed significant discrimination between high- and low-risk patients (p-value < 0.005). This supports that readers’ level of training and clinical experience may not significantly influence the ability to extract accurate radiomic features for NSCLC on CT. This potentially allows flexibility in the training required to produce robust prognostic imaging biomarkers for potential clinical translation.

CT Radiomics Model for Predicting the Ki-67 Index of Lung Cancer： An Exploratory Study

Objective

To establish a radiomics signature and a nomogram model based on enhanced CT images to predict the Ki-67 index of lung cancer.

Methods

From January 2014 to December 2018, 282 patients with lung cancer who had undergone enhanced CT scans and Ki-67 examination within 2 weeks were retrospectively enrolled and analyzed. The clinical data of the patients were collected, such as age, sex, smoking history, maximum tumor diameter and serum tumor markers. Our primary cohort was randomly divided into a training group (n=197) and a validation group (n=85) at a 7:3 ratio. A Ki-67 index ≤ 40% indicated low expression, and a Ki-67 index > 40% indicated high expression. In total, 396 radiomics features were extracted using AK software. Feature reduction and selection were performed using the lasso regression model. Logistic regression analysis was used to establish a multivariate predictive model to identify high and low Ki-67 expression in lung cancer. A nomogram integrating the radiomics score was established based on multiple logistic regression analysis. Area under the curve (AUC) was used to evaluate the prediction efficiency of the radiomics signature and nomogram.

Results

The AUC,sensitivity, specificity and accuracy of the radiomics signature in the training and validation groups were 0.88 (95% CI: 0.82~0.93),79.2%,84.3%,81.2% and 0.86 (95% CI: 0.78~0.94),74.6%,88.1%,79.8%, respectively. A nomogram combining radiomics features and clinical risk factors (smoking history and NSE) was developed. The AUC, sensitivity, specificity and accuracy were 0.87 (95% CI: 0.80~0.95), 75.0%, 90.2% and 83.5% in the validation group, respectively.

Conclusion

The radiomics signature and nomogram based on enhanced CT images provide a way to predict the Ki-67 expression level in lung cancer.

A deep learning- and CT image-based prognostic model for the prediction of survival in non-small cell lung cancer

OBJECTIVE

To assist clinicians in arranging personalized treatment, planning follow-up programs and extending survival times for non-small cell lung cancer (NSCLC) patients, a method of deep learning combined with computed tomography (CT) imaging for survival prediction was designed.

METHODS

Data were collected from 484 patients from four research centers. The data from 344 patients were utilized to build the A_CNN survival prognosis model to classify 2-year overall survival time ranges (730 days cut-off). Data from 140 patients, including independent internal and external test sets, were utilized for model testing. First, a series of preprocessing techniques were used to process the original CT images and generate training and test data sets from the axial, coronal, and sagittal planes. Second, the structure of the A_CNN model was designed based on asymmetric convolution, bottleneck blocks, the uniform cross-entropy (UC) loss function, and other advanced techniques. After that, the A_CNN model was trained, and numerous comparative experiments were designed to obtain the best prognostic survival model. Last, the model performance was evaluated, and the predicted survival curves were analyzed.

RESULTS

The A_CNN survival prognosis model yielded a high patient-level accuracy of 88.8%, a patch-level accuracy of 82.9%, and an area under the receiver operating characteristic (ROC) curve (AUC) of 0.932. When tested on an external data set, the maximum patient-level accuracy was 80.0%.

CONCLUSIONS

The results suggest that using a deep learning method can improve prognosis in patients with NSCLC and has important application value in establishing individualized prognostic models.

Texture Analysis and its Applications in Biomedical Imaging: A Survey

Texture analysis describes a variety of image analysis techniques that quantify the variation in intensity and pattern. This paper provides an overview of several texture analysis approaches addressing the rationale supporting them, their advantages, drawbacks, and applications. This surveys emphasis is in collecting and categorising over five decades of active research on texture analysis. Brief descriptions of different approaches are presented along with application examples. From a broad range of texture analysis applications, this surveys final focus is on biomedical image analysis. An up-to-date list of biological tissues and organs in which disorders produce texture changes that may be used to spot disease onset and progression is provided. Finally, the role of texture analysis methods as biomarkers of disease is summarised.

Strategies to develop radiomics and machine learning models for lung cancer stage and histology prediction using small data samples

Predictive models based on radiomics and machine-learning (ML) need large and annotated datasets for training, often difficult to collect. We designed an operative pipeline for model training to exploit data already available to the scientific community. The aim of this work was to explore the capability of radiomic features in predicting tumor histology and stage in patients with non-small cell lung cancer (NSCLC). We analyzed the radiotherapy planning thoracic CT scans of a proprietary sample of 47 subjects (L-RT) and integrated this dataset with a publicly available set of 130 patients from the MAASTRO NSCLC collection (Lung1). We implemented intra- and inter-sample cross-validation strategies (CV) for evaluating the ML predictive model performances with not so large datasets. We carried out two classification tasks: histology classification (3 classes) and overall stage classification (two classes: stage I and II). In the first task, the best performance was obtained by a Random Forest classifier, once the analysis has been restricted to stage I and II tumors of the Lung1 and L-RT merged dataset (AUC = 0.72 ± 0.11). For the overall stage classification, the best results were obtained when training on Lung1 and testing of L-RT dataset (AUC = 0.72 ± 0.04 for Random Forest and AUC = 0.84 ± 0.03 for linear-kernel Support Vector Machine). According to the classification task to be accomplished and to the heterogeneity of the available dataset(s), different CV strategies have to be explored and compared to make a robust assessment of the potential of a predictive model based on radiomics and ML.

Radiomics-Based Texture Analysis of 68Ga-DOTATATE Positron Emission Tomography and Computed Tomography Images as a Prognostic Biomarker in Adults With Neuroendocrine Cancers Treated With 177Lu-DOTATATE

Purpose

Neuroendocrine tumors (NET) are rare cancers with variable behavior. A better understanding of prognosis would aid individualized management. The aim of this hypothesis-generating pilot study was to investigate the prognostic potential of tumor heterogeneity and tracer avidity in NET using texture analysis (TA) of ⁶⁸Ga-DOTATATE positron emission tomography (PET) and non-enhanced computed tomography (CT) performed at baseline in patients treated with ¹⁷⁷Lu-DOTATATE. It aims to justify a larger-scale study to evaluate its clinical value.

Methods

The pretherapy ⁶⁸Ga-DOTATATE PET-CT scans of 44 patients with metastatic NET (carcinoid, pancreatic, thyroid, head and neck, catecholamine-secreting, and unknown primary NET) treated with ¹⁷⁷Lu-DOTATATE were analyzed retrospectively using commercially available texture analysis research software. Image filtration extracted and enhanced objects of different sizes (fine, medium, coarse), then quantified heterogeneity by statistical and histogram-based parameters (mean intensity, standard deviation, entropy, mean of positive pixels, skewness, and kurtosis). Regions of interest were manually drawn around up to five of the most ⁶⁸Ga-DOTATATE avid lesions for each patient. ⁶⁸Gallium uptake on PET was quantified as SUV_max and SUV_mean. Associations between imaging and clinical markers with progression-free (PFS) and overall survival (OS) were assessed using univariate Kaplan-Meier analysis. Independence of the significant univariate markers of survival was tested using multivariate Cox regression analysis.

Results

Measures of heterogeneity (higher kurtosis, higher entropy, and lower skewness) on coarse-texture scale CT and unfiltered PET images predicted shorter PFS (CT coarse kurtosis: p=0.05, PET entropy: p=0.01, PET skewness: p=0.03) and shorter OS (CT coarse kurtosis: p=0.05, PET entropy: p=0.01, PET skewness p=0.02). Conventional PET parameters such as SUV_max and SUV_mean showed trends towards predicting outcome but were not statistically significant. Multivariate analysis identified that CT-TA (coarse kurtosis: HR=2.57, 95% CI=1.22–5.38, p=0.013) independently predicted PFS, and PET-TA (unfiltered skewness: HR=9.05, 95% CI=1.19–68.91, p=0.033) independently predicted OS.

Conclusion

These preliminary data generate a hypothesis that radiomic analysis of neuroendocrine cancer on ⁶⁸Ga-DOTATATE PET-CT may be of prognostic value and a valuable addition to the assessment of patients.

Chest CT texture-based radiomics analysis in differentiating COVID-19 from other interstitial pneumonia

PURPOSE

To evaluate the potential role of texture-based radiomics analysis in differentiating Coronavirus Disease-19 (COVID-19) pneumonia from pneumonia of other etiology on Chest CT.

MATERIALS AND METHODS

One hundred and twenty consecutive patients admitted to Emergency Department, from March 8, 2020, to April 25, 2020, with suspicious of COVID-19 that underwent Chest CT, were retrospectively analyzed. All patients presented CT findings indicative for interstitial pneumonia. Sixty patients with positive COVID-19 real-time reverse transcription polymerase chain reaction (RT-PCR) and 60 patients with negative COVID-19 RT-PCR were enrolled. CT texture analysis (CTTA) was manually performed using dedicated software by two radiologists in consensus and textural features on filtered and unfiltered images were extracted as follows: mean intensity, standard deviation (SD), entropy, mean of positive pixels (MPP), skewness, and kurtosis. Nonparametric Mann-Whitney test assessed CTTA ability to differentiate positive from negative COVID-19 patients. Diagnostic criteria were obtained from receiver operating characteristic (ROC) curves.

RESULTS

Unfiltered CTTA showed lower values of mean intensity, MPP, and kurtosis in COVID-19 positive patients compared to negative patients (p = 0.041, 0.004, and 0.002, respectively). On filtered images, fine and medium texture scales were significant differentiators; fine texture scale being most significant where COVID-19 positive patients had lower SD (p = 0.004) and MPP (p = 0.004) compared to COVID-19 negative patients. A combination of the significant texture features could identify the patients with positive COVID-19 from negative COVID-19 with a sensitivity of 60% and specificity of 80% (p = 0.001).

CONCLUSIONS

Preliminary evaluation suggests potential role of CTTA in distinguishing COVID-19 pneumonia from other interstitial pneumonia on Chest CT.

Radiomics and functional imaging in lung cancer: the importance of radiological heterogeneity beyond FDG PET/CT and lung biopsy

PURPOSE

[¹⁸F]-Fluorodeoxyglucose Positron Emission Tomography-Computed Tomography (FDG PET/CT) has a central role in the lung nodules' characterization even if, with SUV < 2.5, percutaneous CT-guided Lung Biopsy (CTLB) is needed to assess nodule nature. In that scenario, CT Texture Analysis (CTTA) could be a non-invasive imaging biomarker. Our purpose is to test CTTA ability in differentiating malignant from benign nodules.

METHOD

Patients that underwent FDG PET/CT followed by CTLB between January 2013 and December 2018 were retrospectively enrolled. Were included patients with lung nodule SUV < 2.5 and histological diagnosis.

EXCLUSION CRITERIA

nodules SUV > 2.5, patients who refused CTLB or received oncological treatment before CTLB, indeterminate pathology report, CT motion artifacts. Two radiologists in consensus performed CTTA, drawing a volumetric Region of Interest of nodule with a dedicated first order TA software with and without spatial scaling filters, on preliminary CT performed for CTLB. Statistics included a comparison between malignant and benign neoplasms distribution (2-tailed T-test or Mann-Whitney test according to normal/non-normal data distribution), P-values < 0.05 were considered statistically significant. CTTA accuracy was tested with Receiver Operating Characteristics (ROC) curve.

RESULTS

Form an initial population of 1178, 46 patients encountered inclusion criteria. Pathologist reported 27/46 (59%) malignant and 19/46 (41%) benign nodules. In malignant lesions CTTA showed lower Kurtosis' and higher Skewness' values (all P ≤ 0.0013 and all filtered TA P < 0.024, respectively). ROC curve showed significant Area Under the Curve for Kurtosis and Skewness (0.654 and 0.642, P < 0.001) at medium filtration.

CONCLUSIONS

CTTA is a promising radiological tool to characterize benign and malignant lung nodules, even in those cases without an altered glucose metabolism.

Retinal OCT Texture Analysis for Differentiating Healthy Controls from Multiple Sclerosis (MS) with/without Optic Neuritis

Multiple sclerosis (MS) is an inflammatory disease damaging the myelin sheath in the central and peripheral nervous system in the brain and spinal cord. Optic Neuritis (ON) is one of the most prevalent ocular demonstrations of MS. The current diagnosis protocol for MS is MRI, but newer modalities like Optical Coherence Tomography (OCT) are already of interest in early detection and progression analysis. OCT reveals the symptoms of MS in the Central Nervous System (CNS) through cross-sectional images from neural retinal layers. Previous works on OCT were mostly focused on the thickness of retinal layers; however, texture features seem also to have information in this regard. In this research, we introduce a new pipeline that constructs layer-stacked (LS) images containing data from each specific layer. A variety of texture features are then extracted from LS images to differentiate between healthy controls and ON/None-ON MS cases. Furthermore, the definition of texture extraction methods is tailored for this application. After performing a vast survey on available texture analysis methods, a treasury of powerful features is collected in this paper. As a primary work, this paper shows the ability of such features in the diagnosis of HC and MS (ON and None-ON) cases. Our findings show that the texture features are powerful to diagnose MS cases. Furthermore, adding information of conventional thickness values to texture features improves considerably the discrimination between most of the target groups including HC vs. MS, HC vs. MS-None-ON, and HC vs. MS-ON.

Rethinking a Non-Predominant Pattern in Invasive Lung Adenocarcinoma: Prognostic Dissection Focusing on a High-Grade Pattern

Simple Summary

Prognostic considerations for non-predominant histologic patterns are necessary because most lung adenocarcinomas have a mixed histologic pattern. We aimed to identify prognostic stratification by second most predominant pattern of lung adenocarcinomas and to more accurately assess prognostic factors with CT imaging analysis, particularly enhancing non-predominant but high-grade pattern. We confirmed that the second most predominant histologic pattern can stratify lung adenocarcinoma patients according to prognosis. Especially, when the second most predominant pattern was high-grade, recurrence risk increased by 4.2-fold compared with the low-grade group. Thus, predicting the malignant potential and establishing treatment policies should not rely only on the most predominant pattern. Also, imaging parameters of higher non-contrast CT value and higher SUVmax value are associated with non-predominant but high-grade histologic pattern.

Abstract

Background: Prognostic considerations for non-predominant patterns are necessary because most lung adenocarcinomas (ADCs) have a mixed histologic pattern, and the spectrum of actual prognosis varies widely even among lung ADCs with the same most predominant pattern. We aimed to identify prognostic stratification by second most predominant pattern of lung ADC and to more accurately assess prognostic factors with CT imaging analysis, particularly enhancing non-predominant but high-grade pattern. Methods: In this prospective study, patients with early-stage lung ADC undergoing curative surgery underwent preoperative dual-energy CT (DECT) and positron emission tomography (PET)/CT. Histopathology of ADC, the most predominant and second most predominant histologic patterns, and preoperative imaging parameters were assessed and correlated with patient survival. Results: Among the 290 lung ADCs included in the study, 231 (79.7%) were mixed-pathologic pattern. When the most predominant histologic pattern was intermediate-grade, survival curves were significantly different among the three second most predominant subgroups (p = 0.004; low, lepidic; intermediate, acinar and papillary; high, micropapillary and solid). When the second most predominant pattern was high-grade, recurrence risk increased by 4.2-fold compared with the low-grade group (p = 0.005). To predict a non-predominant but high-grade pattern, the non-contrast CT value of tumor was meaningful with a lower HU value associated with the histologic combination of lower grade (low-grade as most predominant and intermediate-grade as second most predominant pattern, OR = 6.15, p = 0.005; intermediate-grade as most predominant and high-grade as second most predominant pattern, OR = 0.10, p = 0.033). SUVmax of the tumor was associated with the non-predominant but high-grade pattern, especially in the histologic combination of intermediate-high grade (OR = 1.14, p = 0.012). Conclusions: The second most predominant histologic pattern can stratify lung ADC patients according to prognosis. Thus, predicting the malignant potential and establishing treatment policies should not rely only on the most predominant pattern. Moreover, imaging parameters of non-contrast CT value and SUVmax could be useful in predicting a non-predominant but high-grade histologic pattern.

Quantification of Structural Heterogeneity Using Fractal Analysis of Contrast-Enhanced CT Image to Predict Survival in Gastric Cancer Patients

BACKGROUND

Malignant tumor essentially implies structural heterogeneity. Fractal analysis of medical imaging has a potential to quantify this structural heterogeneity in the tumor AIMS: The purpose of this study is to quantify this structural abnormality in the tumor applying fractal analysis to contrast-enhanced computed tomography (CE-CT) image and to evaluate its biomarker value for predicting survival of surgically treated gastric cancer patients.

METHODS

A total of 108 gastric cancer patients (77 men and 31 women; mean age: 69.1 years), who received curative surgery without any neoadjuvant therapy, were retrospectively investigated. Portal-phase CE-CT images were analyzed with use of a plug-in tool for ImageJ (NIH, Bethesda, USA), and the fractal dimension (FD) in the tumor was calculated using a differential box-counting method to quantify structural heterogeneity in the tumor. Tumor FD was compared with clinicopathologic features and disease-specific survival (DSS).

RESULTS

High FD value of the tumor significantly associated with high T stage and high pathological stage (P = 0.009, 0.007, respectively). In Kaplan-Meier analysis, patients with higher FD tumors (FD > 0.9746) showed a significantly worse DSS (P = 0.009, log rank). Multivariate analysis demonstrated that tumor FD, T stage, and N stage were independent prognostic factors for DSS. In subset analysis of lymph-node positive gastric cancers, only tumor FD was an independent prognostic factor for DSS.

CONCLUSION

CT fractal analysis can be a useful biomarker for gastric cancer patients, reflecting survival and clinicopathologic features.

Texture Analysis of Fractional Water Content Images Acquired during PET/MRI: Initial Evidence for an Association with Total Lesion Glycolysis, Survival and Gene Mutation Profile in Primary Colorectal Cancer

To assess the capability of fractional water content (FWC) texture analysis (TA) to generate biologically relevant information from routine PET/MRI acquisitions for colorectal cancer (CRC) patients. Thirty consecutive primary CRC patients (mean age 63.9, range 42-83 years) prospectively underwent FDG-PET/MRI. FWC tumor parametric images generated from Dixon MR sequences underwent TA using commercially available research software (TexRAD). Data analysis comprised (1) identification of functional imaging correlates for texture features (TF) with low inter-observer variability (intraclass correlation coefficient: ICC > 0.75), (2) evaluation of prognostic performance for FWC-TF, and (3) correlation of prognostic imaging signatures with gene mutation (GM) profile. Of 32 FWC-TF with ICC > 0.75, 18 correlated with total lesion glycolysis (TLG, highest: rs = -0.547, p = 0.002). Using optimized cut-off values, five MR FWC-TF identified a good prognostic group with zero mortality (lowest: p = 0.017). For the most statistically significant prognostic marker, favorable prognosis was significantly associated with a higher number of GM per patient (medians: 7 vs. 1.5, p = 0.009). FWC-TA derived from routine PET/MRI Dixon acquisitions shows good inter-operator agreement, generates biological relevant information related to TLG, GM count, and provides prognostic information that can unlock new clinical applications for CRC patients.

Optimising use of 4D-CT phase information for radiomics analysis in lung cancer patients treated with stereotactic body radiotherapy

Purpose. 4D-CT is routine imaging for lung cancer patients treated with stereotactic body radiotherapy. No studies have investigated optimal 4D phase selection for radiomics. We aim to determine how phase data should be used to identify prognostic biomarkers for distant failure, and test whether stability assessment is required. A phase selection approach will be developed to aid studies with different 4D protocols and account for patient differences. Methods. 186 features were extracted from the tumour and peritumour on all phases for 258 patients. Feature values were selected from phase features using four methods: (A) mean across phases, (B) median across phases, (C) 50% phase, and (D) the most stable phase (closest in value to two neighbours), coined personalised selection. Four levels of stability assessment were also analysed, with inclusion of: (1) all features, (2) stable features across all phases, (3) stable features across phase and neighbour phases, and (4) features averaged over neighbour phases. Clinical-radiomics models were built for twelve combinations of feature type and assessment method. Model performance was assessed by concordance index (c-index) and fraction of new information from radiomic features. Results. The most stable phase spanned the whole range but was most often near exhale. All radiomic signatures provided new information for distant failure prediction. The personalised model had the highest c-index (0.77), and 58% of new information was provided by radiomic features when no stability assessment was performed. Conclusion. The most stable phase varies per-patient and selecting this improves model performance compared to standard methods. We advise the single most stable phase should be determined by minimising feature differences to neighbour phases. Stability assessment over all phases decreases performance by excessively removing features. Instead, averaging of neighbour phases should be used when stability is of concern. The models suggest that higher peritumoural intensity predicts distant failure.

Treatment response prediction using MRI-based pre-, post-, and delta-radiomic features and machine learning algorithms in colorectal cancer

OBJECTIVES

We evaluate the feasibility of treatment response prediction using MRI-based pre-, post-, and delta-radiomic features for locally advanced rectal cancer (LARC) patients treated by neoadjuvant chemoradiation therapy (nCRT).

MATERIALS AND METHODS

This retrospective study included 53 LARC patients divided into a training set (Center#1, n = 36) and external validation set (Center#2, n = 17). T2-weighted (T2W) MRI was acquired for all patients, 2 weeks before and 4 weeks after nCRT. Ninety-six radiomic features, including intensity, morphological and second- and high-order texture features were extracted from segmented 3D volumes from T2W MRI. All features were harmonized using ComBat algorithm. Max-Relevance-Min-Redundancy (MRMR) algorithm was used as feature selector and k-nearest neighbors (KNN), Naïve Bayes (NB), Random forests (RF), and eXtreme Gradient Boosting (XGB) algorithms were used as classifiers. The evaluation was performed using the area under the receiver operator characteristic (ROC) curve (AUC), sensitivity, specificity and accuracy.

RESULTS

In univariate analysis, the highest AUC in pre-, post-, and delta-radiomic features were 0.78, 0.70, and 0.71, for GLCM_IMC1, shape (surface area and volume) and GLSZM_GLNU features, respectively. In multivariate analysis, RF and KNN achieved the highest AUC (0.85 ± 0.04 and 0.81 ± 0.14, respectively) among pre- and post-treatment features. The highest AUC was achieved for the delta-radiomic-based RF model (0.96 ± 0.01) followed by NB (0.96 ± 0.04). Overall. Delta-radiomics model, outperformed both pre- and post-treatment features (P-value <0.05).

CONCLUSION

Multivariate analysis of delta-radiomic T2W MRI features using machine learning algorithms could potentially be used for response prediction in LARC patients undergoing nCRT. We also observed that multivariate analysis of delta-radiomic features using RF classifiers can be used as powerful biomarkers for response prediction in LARC.

Radiomics-Based Features for Prediction of Histological Subtypes in Central Lung Cancer

Objectives

To evaluate the effectiveness of radiomic features on classifying histological subtypes of central lung cancer in contrast-enhanced CT (CECT) images.

Materials and Methods

A total of 200 patients with radiologically defined central lung cancer were recruited. All patients underwent dual-phase chest CECT, and the histological subtypes (adenocarcinoma (ADC), squamous cell carcinoma (SCC), small cell lung cancer (SCLC)) were confirmed by histopathological samples. 107 features were used in five machine learning classifiers to perform the predictive analysis among three subtypes. Models were trained and validated in two conditions: using radiomic features alone, and combining clinical features with radiomic features. The performance of the classification models was evaluated by the area under the receiver operating characteristic curve (AUC).

Results

The highest AUCs in classifying ADC vs. SCC, ADC vs. SCLC, and SCC vs. SCLC were 0.879, 0.836, 0.783, respectively by using only radiomic features in a feedforward neural network.

Conclusion

Our study indicates that radiomic features based on the CECT images might be a promising tool for noninvasive prediction of histological subtypes in central lung cancer and the neural network classifier might be well-suited to this task.

Comparison of CT Texture Analysis Software Platforms in Renal Cell Carcinoma: Reproducibility of Numerical Values and Association With Histologic Subtype Across Platforms

OBJECTIVE. The purpose of this article is to evaluate interobserver, intraobserver, and interplatform variability and compare the previously established association between texture metrics and tumor histologic subtype using three commercially available CT texture analysis (CTTA) software platforms on the same dataset of large (> 7 cm) renal cell carcinomas (RCCs). MATERIALS AND METHODS. CT-based texture analysis was performed on contrast-enhanced MDCT images of large (> 7 cm) untreated RCCs in 124 patients (median age, 62 years; 82 men and 42 women) using three different software platforms. Using this previously studied cohort, texture features were compared across platforms. Features were correlated with histologic subtype, and strength of association was compared between platforms. Single-slice and volumetric measures from one platform were compared. Values for interobserver and intraobserver variability on a tumor subset (n = 30) were assessed across platforms. RESULTS. Metrics including mean gray-level intensity, SD, and volume correlated fairly well across platforms (concordance correlation coefficient [CCC], 0.66-0.99; mean relative difference [MRD], 0.17-5.97%). Entropy showed high variability (CCC, 0.04; MRD, 44.5%). Mean, SD, mean of positive pixels (MPP), and entropy were associated with clear cell histologic subtype on almost all platforms (p < .05). Mean, SD, entropy, and MPP were highly reproducible on most platforms on both interobserver and intraobserver analysis. CONCLUSION. Select texture metrics were reproducible across platforms and readers, but other metrics were widely variable. If clinical models are developed that use CTTA for medical decision making, these differences in reproducibility of some features across platforms need to be considered, and standardization is critical for more widespread adaptation and implementation.

Prognostic Modeling of Patients Undergoing Surgery Alone for Esophageal Squamous Cell Carcinoma: A Histopathological and Computed Tomography Based Quantitative Analysis

Objective

To evaluate the effectiveness of a novel computerized quantitative analysis based on histopathological and computed tomography (CT) images for predicting the postoperative prognosis of esophageal squamous cell carcinoma (ESCC) patients.

Methods

We retrospectively reviewed the medical records of 153 ESCC patients who underwent esophagectomy alone and quantitatively analyzed digital histological specimens and diagnostic CT images. We cut pathological images (6000 × 6000) into 50 × 50 patches; each patient had 14,400 patches. Cluster analysis was used to process these patches. We used the pathological clusters to all patches ratio (PCPR) of each case for pathological features and we obtained 20 PCPR quantitative features. Totally, 125 computerized quantitative (20 PCPR and 105 CT) features were extracted. We used a recursive feature elimination approach to select features. A Cox hazard model with L1 penalization was used for prognostic indexing. We compared the following prognostic models: Model A: clinical features; Model B: quantitative CT and clinical features; Model C: quantitative histopathological and clinical features; and Model D: combined information of clinical, CT, and histopathology. Indices of concordance (C-index) and leave-one-out cross-validation (LOOCV) were used to assess prognostic model accuracy.

Results

Five PCPR and eight CT features were treated as significant indicators in ESCC prognosis. C-indices adjusted for LOOCV were comparable among four models, 0.596 (Model A) vs. 0.658 (Model B) vs. 0.651 (Model C), and improved to 0.711with Model D combining information of clinical, CT, and histopathology (all p<0.05). Using Model D, we stratified patients into low- and high-risk groups. The 3-year overall survival rates of low- and high-risk patients were 38.0% and 25.0%, respectively (p<0.001).

Conclusion

Quantitative prognostic modeling using a combination of clinical data, histopathological, and CT images can stratify ESCC patients with surgery alone into high-risk and low-risk groups.

Combined radiomics-clinical model to predict malignancy of vertebral compression fractures on CT

OBJECTIVES

To develop and validate a combined radiomics-clinical model to predict malignancy of vertebral compression fractures on CT.

METHODS

One hundred sixty-five patients with vertebral compression fractures were allocated to training (n = 110 [62 acute benign and 48 malignant fractures]) and validation (n = 55 [30 acute benign and 25 malignant fractures]) cohorts. Radiomics features (n = 144) were extracted from non-contrast-enhanced CT images. Radiomics score was constructed by applying least absolute shrinkage and selection operator regression to reproducible features. A combined radiomics-clinical model was constructed by integrating significant clinical parameters with radiomics score using multivariate logistic regression analysis. Model performance was quantified in terms of discrimination and calibration. The model was internally validated on the independent data set.

RESULTS

The combined radiomics-clinical model, composed of two significant clinical predictors (age and history of malignancy) and the radiomics score, showed good calibration (Hosmer-Lemeshow test, p > 0.05) and discrimination in both training (AUC, 0.970) and validation (AUC, 0.948) cohorts. Discrimination performance of the combined model was higher than that of either the radiomics score (AUC, 0.941 in training cohort and 0.852 in validation cohort) or the clinical predictor model (AUC, 0.924 in training cohort and 0.849 in validation cohort). The model stratified patients into groups with low and high risk of malignant fracture with an accuracy of 98.2% in the training cohort and 90.9% in the validation cohort.

CONCLUSIONS

The combined radiomics-clinical model integrating clinical parameters with radiomics score could predict malignancy in vertebral compression fractures on CT with high discriminatory ability.

KEY POINTS

• A combined radiomics-clinical model was constructed to predict malignancy of vertebral compression fractures on CT by combining clinical parameters and radiomics features. • The model showed good calibration and discrimination in both training and validation cohorts. • The model showed high accuracy in the stratification of patients into groups with low and high risk of malignant vertebral compression fractures.

Deep learning classification of lung cancer histology using CT images

Tumor histology is an important predictor of therapeutic response and outcomes in lung cancer. Tissue sampling for pathologist review is the most reliable method for histology classification, however, recent advances in deep learning for medical image analysis allude to the utility of radiologic data in further describing disease characteristics and for risk stratification. In this study, we propose a radiomics approach to predicting non-small cell lung cancer (NSCLC) tumor histology from non-invasive standard-of-care computed tomography (CT) data. We trained and validated convolutional neural networks (CNNs) on a dataset comprising 311 early-stage NSCLC patients receiving surgical treatment at Massachusetts General Hospital (MGH), with a focus on the two most common histological types: adenocarcinoma (ADC) and Squamous Cell Carcinoma (SCC). The CNNs were able to predict tumor histology with an AUC of 0.71(p = 0.018). We also found that using machine learning classifiers such as k-nearest neighbors (kNN) and support vector machine (SVM) on CNN-derived quantitative radiomics features yielded comparable discriminative performance, with AUC of up to 0.71 (p = 0.017). Our best performing CNN functioned as a robust probabilistic classifier in heterogeneous test sets, with qualitatively interpretable visual explanations to its predictions. Deep learning based radiomics can identify histological phenotypes in lung cancer. It has the potential to augment existing approaches and serve as a corrective aid for diagnosticians.

Machine and deep learning methods for radiomics

Radiomics is an emerging area in quantitative image analysis that aims to relate large-scale extracted imaging information to clinical and biological endpoints. The development of quantitative imaging methods along with machine learning has enabled the opportunity to move data science research towards translation for more personalized cancer treatments. Accumulating evidence has indeed demonstrated that noninvasive advanced imaging analytics, that is, radiomics, can reveal key components of tumor phenotype for multiple three-dimensional lesions at multiple time points over and beyond the course of treatment. These developments in the use of CT, PET, US, and MR imaging could augment patient stratification and prognostication buttressing emerging targeted therapeutic approaches. In recent years, deep learning architectures have demonstrated their tremendous potential for image segmentation, reconstruction, recognition, and classification. Many powerful open-source and commercial platforms are currently available to embark in new research areas of radiomics. Quantitative imaging research, however, is complex and key statistical principles should be followed to realize its full potential. The field of radiomics, in particular, requires a renewed focus on optimal study design/reporting practices and standardization of image acquisition, feature calculation, and rigorous statistical analysis for the field to move forward. In this article, the role of machine and deep learning as a major computational vehicle for advanced model building of radiomics-based signatures or classifiers, and diverse clinical applications, working principles, research opportunities, and available computational platforms for radiomics will be reviewed with examples drawn primarily from oncology. We also address issues related to common applications in medical physics, such as standardization, feature extraction, model building, and validation.

Quantitative tumor heterogeneity MRI profiling improves machine learning-based prognostication in patients with metastatic colon cancer

OBJECTIVES

Intra-tumor heterogeneity has been previously shown to be an independent predictor of patient survival. The goal of this study is to assess the role of quantitative MRI-based measures of intra-tumor heterogeneity as predictors of survival in patients with metastatic colorectal cancer.

METHODS

In this IRB-approved retrospective study, we identified 55 patients with stage 4 colon cancer with known hepatic metastasis on MRI. Ninety-four metastatic hepatic lesions were identified on post-contrast images and manually volumetrically segmented. A heterogeneity phenotype vector was extracted from each lesion. Univariate regression analysis was used to assess the contribution of 110 extracted features to survival prediction. A random forest-based machine learning technique was applied to the feature vector and to the standard prognostic clinical and pathologic variables. The dataset was divided into a training and test set at a ratio of 4:1. ROC analysis and confusion matrix analysis were used to assess classification performance.

RESULTS

Mean survival time was 39 ± 3.9 months for the study population. A total of 22 texture features were associated with patient survival (p < 0.05). The trained random forest machine learning model that included standard clinical and pathological prognostic variables resulted in an area under the ROC curve of 0.83. A model that adds imaging-based heterogeneity features to the clinical and pathological variables resulted in improved model performance for survival prediction with an AUC of 0.94.

CONCLUSIONS

MRI-based texture features are associated with patient outcomes and improve the performance of standard clinical and pathological variables for predicting patient survival in metastatic colorectal cancer.

KEY POINTS

• MRI-based tumor heterogeneity texture features are associated with patient survival outcomes. • MRI-based tumor texture features complement standard clinical and pathological variables for prognosis prediction in metastatic colorectal cancer. • Agglomerative hierarchical clustering shows that patient survival outcomes are associated with different MRI tumor profiles.

Stage-Specific PET Radiomic Prediction Model for the Histological Subtype Classification of Non-Small-Cell Lung Cancer

Purpose

To investigate the impact of staging on differences in glucose metabolic heterogeneity between lung adenocarcinoma (ADC) and squamous cell carcinoma (SCC) by ¹⁸F-fluorodeoxyglucose positron emission tomography (¹⁸F-FDG PET) textural analysis and to develop a stage-specific PET radiomic prediction model to distinguish lung ADC from SCC.

Patients and Methods

Patients who were histologically diagnosed with lung ADC or SCC and underwent pretreatment ¹⁸F-FDG PET/CT scans were retrospectively identified. Radiomic features were extracted from a semiautomatically outlined tumor region in the Chang-Gung Image Texture Analysis (CGITA) software package. The differences in radiomic parameters between lung ADC and SCC were compared stage-by-stage in 253 consecutive NSCLC patients with stages I to III disease. The least absolute shrinkage and selection operator (LASSO) algorithm was used for feature selection. A radiomic signature for each stage was subsequently constructed and evaluated. Then, an individual nomogram incorporating the radiomic signature and clinical risk factors was established and evaluated. The performance of the constructed models was assessed by receiver operating characteristic (ROC) curve analysis, and the nomogram was further validated by calibration curve analysis.

Results

The performance of the radiomic signature for distinguishing lung ADC and SCC in both the training and validation cohorts was good, with AUCs of 0.883, 0.854, and 0.895 in the training cohort and 0.932, 0.944, and 0.886 in the validation cohort for stages I, II, and III NSCLC, respectively. The radiomic-clinical nomogram integrating radiomic features with independent clinical predictors exhibited more favorable discriminative performance, with AUCs of 0.982, 0.963, and 0.979 in the training cohort and 0.989, 0.984, and 0.978 in the validation cohort for stages I, II, and III, respectively.

Conclusion

Differences in PET radiomic features between lung ADC and SCC varied in different stages. Stage-specific PET radiomic prediction models provided more favorable performance for discriminating the histological subtype of NSCLC.

Preoperative Predicting the WHO/ISUP Nuclear Grade of Clear Cell Renal Cell Carcinoma by Computed Tomography-Based Radiomics Features

Nuclear grade is important for treatment selection and prognosis in patients with clear cell renal cell carcinoma (ccRCC). This study aimed to determine the ability of preoperative four-phase multiphasic multidetector computed tomography (MDCT)-based radiomics features to predict the WHO/ISUP nuclear grade. In all 102 patients with histologically confirmed ccRCC, the training set (n = 62) and validation set (n = 40) were randomly assigned. In both datasets, patients were categorized according to the WHO/ISUP grading system into low-grade ccRCC (grades 1 and 2) and high-grade ccRCC (grades 3 and 4). The feature selection process consisted of three steps, including least absolute shrinkage and selection operator (LASSO) regression analysis, and the radiomics scores were developed using 48 radiomics features (10 in the unenhanced phase, 17 in the corticomedullary (CM) phase, 14 in the nephrographic (NP) phase, and 7 in the excretory phase). The radiomics score (Rad-Score) derived from the CM phase achieved the best predictive ability, with a sensitivity, specificity, and an area under the curve (AUC) of 90.91%, 95.00%, and 0.97 in the training set. In the validation set, the Rad-Score derived from the NP phase achieved the best predictive ability, with a sensitivity, specificity, and an AUC of 72.73%, 85.30%, and 0.84. We constructed a complex model, adding the radiomics score for each of the phases to the clinicoradiological characteristics, and found significantly better performance in the discrimination of the nuclear grades of ccRCCs in all MDCT phases. The highest AUC of 0.99 (95% CI, 0.92-1.00, p < 0.0001) was demonstrated for the CM phase. Our results showed that the MDCT radiomics features may play a role as potential imaging biomarkers to preoperatively predict the WHO/ISUP grade of ccRCCs.

Artificial intelligence and hybrid imaging: the best match for personalized medicine in oncology

Artificial intelligence (AI) refers to a field of computer science aimed to perform tasks typically requiring human intelligence. Currently, AI is recognized in the broader technology radar within the five key technologies which emerge for their wide-ranging applications and impact in communities, companies, business, and value chain framework alike. However, AI in medical imaging is at an early phase of development, and there are still hurdles to take related to reliability, user confidence, and adoption. The present narrative review aimed to provide an overview on AI-based approaches (distributed learning, statistical learning, computer-aided diagnosis and detection systems, fully automated image analysis tool, natural language processing) in oncological hybrid medical imaging with respect to clinical tasks (detection, contouring and segmentation, prediction of histology and tumor stage, prediction of mutational status and molecular therapies targets, prediction of treatment response, and outcome). Particularly, AI-based approaches have been briefly described according to their purpose and, finally lung cancer-being one of the most extensively malignancy studied by hybrid medical imaging-has been used as illustrative scenario. Finally, we discussed clinical challenges and open issues including ethics, validation strategies, effective data-sharing methods, regulatory hurdles, educational resources, and strategy to facilitate the interaction among different stakeholders. Some of the major changes in medical imaging will come from the application of AI to workflow and protocols, eventually resulting in improved patient management and quality of life. Overall, several time-consuming tasks could be automatized. Machine learning algorithms and neural networks will permit sophisticated analysis resulting not only in major improvements in disease characterization through imaging, but also in the integration of multiple-omics data (i.e., derived from pathology, genomic, proteomics, and demographics) for multi-dimensional disease featuring. Nevertheless, to accelerate the transition of the theory to practice a sustainable development plan considering the multi-dimensional interactions between professionals, technology, industry, markets, policy, culture, and civil society directed by a mindset which will allow talents to thrive is necessary.

Texture features from computed tomography correlate with markers of severity in acute alcohol-associated hepatitis

The aim of this study was to use texture analysis to establish quantitative CT-based imaging features to predict clinical severity in patients with acute alcohol-associated hepatitis (AAH). A secondary aim was to compare the performance of texture analysis to deep learning. In this study, mathematical texture features were extracted from CT slices of the liver for 34 patients with a diagnosis of AAH and 35 control patients. Recursive feature elimination using random forest (RFE-RF) was used to identify the best combination of features to distinguish AAH from controls. These features were subsequently used as predictors to determine associated clinical values. To compare machine learning with deep learning approaches, a 2D dense convolutional neural network (CNN) was implemented and trained for the classification task of AAH. RFE-RF identified 23 top features used to classify AAH images, and the subsequent model demonstrated an accuracy of 82.4% in the test set. The deep learning CNN demonstrated an accuracy of 70% in the test set. We show that texture features of the liver are unique in AAH and are candidate quantitative biomarkers that can be used in prospective studies to predict the severity and outcomes of patients with AAH.

Clinical T category for lung cancer staging: A pragmatic approach for real-world practice

BACKGROUND

To determine which components should be measured and which window settings are appropriate for computerized tomography (CT) size measurements of lung adenocarcinoma (ADC) and to explore interobserver agreement and accuracy according to the eighth edition of TNM staging.

METHODS

A total of 165 patients with surgically resected lung ADC earlier than stage 3A were included in this study. One radiologist and two pulmonologists independently measured the total and solid sizes of components of tumors on different window settings and assessed solidity. CT measurements were compared with pathologic size measurements.

RESULTS

In categorizing solidity, 25% of the cases showed discordant results among observers. Measuring the total size of a lung adenocarcinoma predicted pathologic invasive components to a degree similar to measuring the solid component. Lung windows were more accurate (intraclass correlation [ICC] = 0.65-0.81) than mediastinal windows (ICC = 0.20-0.72) at predicting pathologic invasive components, especially in a part-solid nodule. Interobserver agreements for measurement of solid components were good with little significant difference (lung windows, ICC = 0.89; mediastinal windows, ICC = 0.91). A high level of interobserver agreement was seen between the radiologist and pulmonologists and between residents (from the division of pulmonology and critical care) versus a fellow (from the division of pulmonology and critical care) on different windows.

CONCLUSIONS

A considerable percentage (25%) of discrepancies was encountered in categorizing the solidity of lesions, which may decrease the accuracy of measurements. Lung window settings may be superior to mediastinal windows for measuring lung ADCs, with comparable interobserver agreement and moderate accuracy for predicting pathologic invasive components.

KEY POINTS

SIGNIFICANT FINDINGS OF THE STUDY: Lung window settings are better for evaluating part-solid lung adenocarcinoma (ADC), with comparable interobserver agreement and moderate accuracy for predicting pathologic invasive components. The considerable percentage (25%) of discrepancies in categorizing solidity of the lesions may also have decreased the accuracy of measurements.

WHAT THIS STUDY ADDS

For accurate measurement and categorization of lung ADC, robust quantitative analysis is needed rather than a simple visual assessment.