Radiomics and radiogenomics in lung cancer: A review for the clinician

CT Densitometry and Morphology of Radiofrequency-Ablated Stage IA Non-Small Cell Lung Cancer: Results from the American College of Surgeons Oncology Group Z4033 (Alliance) Trial

PURPOSE

To evaluate tumor and ablation zone morphology and densitometry related to tumor recurrence in participants with Stage IA non-small cell lung cancer undergoing radiofrequency ablation in a prospective, multicenter trial.

MATERIALS AND METHODS

Forty-five participants (median 76 years old; 25 women; 20 men) from 16 sites were followed for 2 years (December 2006 to November 2010) with computed tomography (CT) densitometry. Imaging findings before and after ablation were recorded, including maximum CT attenuation (in Hounsfield units) at precontrast and 45-, 90-, 180-, and 300-s postcontrast.

RESULTS

Every 1-cm increase in the largest axial diameter of the ablation zone at 3-months' follow-up compared to the index tumor reduced the odds of 2-year recurrence by 52% (P = .02). A 1-cm difference performed the best (sensitivity, 0.56; specificity, 0.93; positive likelihood ratio of 8). CT densitometry precontrast and at 45 seconds showed significantly different enhancement patterns in a comparison among pretreated lung cancer (delta = +61.2 HU), tumor recurrence (delta = +57 HU), and treated tumor/ablation zone (delta [change in attenuation] = +16.9 HU), (P < .0001). Densitometry from 45 to 300 s was also different among pretreated tumor (delta = -6.8 HU), recurrence (delta = -11.2 HU), and treated tumor (delta = +12.1 HU; P = .01). Untreated and residual tumor demonstrated washout, whereas treated tumor demonstrated increased attenuation.

CONCLUSIONS

An ablation zone ≥1 cm larger than the initial tumor, based on 3-month follow-up imaging, is recommended to decrease odds of recurrence. CT densitometry can delineate tumor versus treatment zones.

Effects of simulated dose variation on contrast-enhanced CT-based radiomic analysis for Non-Small Cell Lung Cancer

PURPOSE

To examine the potential effect of CT dose variation on radiomic features in vivo using simulated contrast-enhanced CT dose reduction in patients with non-small lung cell cancer (NSCLC).

METHODS

In this retrospective study, we included 69 patients (25 females, 44 males, median age 66 years) with histologically proven NSCLC who underwent a whole contrast-enhanced body FDG-PET/CT for primary staging. To simulate different CT dose levels, we used an algorithm to simulate low-dose CT images based on a noise model derived from phantom experiments. The tumor lesions and reference regions in the paraspinal muscle were manually segmented to obtain three-dimensional regions of interest. Radiomic feature extraction was performed using the PyRadiomics toolbox. The median relative feature value deviation was assessed for each feature and each dose level.

RESULTS

The mean segmented tumor volume was 340 ml. T-stages of the primary tumors were primarily T3/4. For NSCLCs, the median relative feature value deviation in the lowest dose images varied for the calculated features from 52.2% to -49.5%. In general, dose-dependent deviations of feature values showed a monotonous increase or decrease with decreasing dose levels. Statistical analyses revealed significant differences between the dose levels in 91% of features.

CONCLUSIONS

We examined the effects of simulated CT dose reduction on the values of radiomic features in primary NSCLC and showed significant deviations of varying degrees in numerous feature classes. This is a theoretical indicator of potential influence under real conditions, which should be taken into account in clinical use.

PD-1-positive Natural Killer Cells have a weaker antitumor function than that of PD-1-negative Natural Killer Cells in Lung Cancer

Antibodies targeting the immune checkpoint inhibitor, programmed cell death 1 (PD-1), have provided a breakthrough in the treatment of lung cancer. However, the function of PD-1 in natural killer (NK) cells of cancer patients remains unclear. Herein, we analyzed the expression of PD-1 on the NK cells in the peripheral blood of patients with lung cancer and found that the level of PD-1⁺ NK cells in patients was significantly higher than that in healthy individuals. Moreover, these PD-1⁺ NK cells demonstrated a weaker ability to secrete interferon-gamma (INF-γ), granzyme B, and perforin, and exhibited lower CD107a expression. Importantly, in patients with lung cancer, the percentage of PD-1⁺ NK cells was significantly positively correlated with the concentration of IL-2 in the plasma, which was also higher than that in healthy individuals. In addition, IL-2 could increase the expression of PD-1 on NK cells in vitro, indicating that high IL-2 level in the plasma is largely responsible for the abundance of PD-1⁺ NK cells in patients with lung cancer. These findings demonstrate intriguing mechanisms for understanding the expression of PD-1 on NK cells and the function of PD-1⁺ NK cells in lung cancer. This study confirms and extends previous studies demonstrating that PD-1 can negatively regulate the antitumor function of NK cells.

Research progress of computer aided diagnosis system for pulmonary nodules in CT images

BACKGROUND AND OBJECTIVE

Since CAD (Computer Aided Diagnosis) system can make it easier and more efficient to interpret CT (Computer Tomography) images, it has gained much attention and developed rapidly in recent years. This article reviews recent CAD techniques for pulmonary nodule detection and diagnosis in CT Images.

METHODS

CAD systems can be classified into computer-aided detection (CADe) and computer-aided diagnosis (CADx) systems. This review reports recent researches of both systems, including the database, technique, innovation and experimental results of each work. Multi-task CAD systems, which can handle segmentation, false positive reduction, malignancy prediction and other tasks at the same time. The commercial CAD systems are also briefly introduced.

RESULTS

We have found that deep learning based CAD is the mainstream of current research. The reported sensitivity of deep learning based CADe systems ranged between 80.06% and 94.1% with an average 4.3 false-positive (FP) per scan when using LIDC-IDRI dataset, and between 94.4% and 97.9% with an average 4 FP/scan when using LUNA16 dataset, respectively. The overall accuracy of deep learning based CADx systems ranged between 86.84% and 92.3% with an average AUC of 0.956 reported when using LIDC-IDRI dataset.

CONCLUSIONS

We summarized the current tendency and limitations as well as future challenges in this field. The development of CAD needs to meet the rigid clinical requirements, such as high accuracy, strong robustness, high efficiency, fine-grained analysis and classification, and to provide practical clinical functions. This review provides helpful information for both engineering researchers and radiologists to learn the latest development of CAD systems.

Assessing EGFR gene mutation status in non-small cell lung cancer with imaging features from PET/CT

OBJECTIVE

The aim of this study was to investigate whether quantitative and qualitative features extracted from PET/computed tomography (CT) can be used as imaging biomarkers for evaluating epidermal growth factor receptor (EGFR) mutation status in non-small cell lung cancer patients.

METHODS

Eighty patients with stage II and III non-small cell lung cancer from January 2017 to December 2017 were included in this study. All patients underwent PET/CT examination before operation. Patients with 30 EGFR positive and 50 EGFR negative were confirmed by pathological verification and gene detection. Least absolute shrinkage and selection operator was used for analysis and selection of imaging features. Support vector machine was used to classify EGFR positive/negative using the selected features. Ten-fold cross validation was used to estimate the accuracy.

RESULTS

A total of 512 quantitative features (radiomic features) were extracted from PET/CT (256 for PET and 256 for CT), and 12 qualitative features (semantic features) were extracted from CT. A total of 35 features were finally retained after least absolute shrinkage and selection operator (31 quantitative features and 4 qualitative features). The 35 selected features were significantly associated with EGFR mutation status. A predictive model was built using PET/CT data. Its performance was revealed as 0.953 using the area under the receiver operating characteristic curve.

CONCLUSION

A predictive model using PET/CT images might be used to detect EGFR mutation status in non-small cell lung cancer patients.

Myocardial Radiomics in Cardiac MRI

OBJECTIVE. The purpose of this article is to review the nascent field of radiomics in cardiac MRI. CONCLUSION. Cardiac MRI produces a large number of images in a fairly inefficient manner with sometimes limited clinical application. In the era of precision medicine, there is increasing need for imaging to account for a broader array of diseases in an efficient and objective manner. Radiomics, the extraction and analysis of quantitative imaging features from medical imaging, may offer potential solutions to this need.

Artificial intelligence, machine learning, computer-aided diagnosis, and radiomics: advances in imaging towards to precision medicine

The discipline of radiology and diagnostic imaging has evolved greatly in recent years. We have observed an exponential increase in the number of exams performed, subspecialization of medical fields, and increases in accuracy of the various imaging methods, making it a challenge for the radiologist to "know everything about all exams and regions". In addition, imaging exams are no longer only qualitative and diagnostic, providing now quantitative information on disease severity, as well as identifying biomarkers of prognosis and treatment response. In view of this, computer-aided diagnosis systems have been developed with the objective of complementing diagnostic imaging and helping the therapeutic decision-making process. With the advent of artificial intelligence, "big data", and machine learning, we are moving toward the rapid expansion of the use of these tools in daily life of physicians, making each patient unique, as well as leading radiology toward the concept of multidisciplinary approach and precision medicine. In this article, we will present the main aspects of the computational tools currently available for analysis of images and the principles of such analysis, together with the main terms and concepts involved, as well as examining the impact that the development of artificial intelligence has had on radiology and diagnostic imaging.

Multiparametric MRI and auto-fixed volume of interest-based radiomics signature for clinically significant peripheral zone prostate cancer

Objectives

To create a radiomics approach based on multiparametric magnetic resonance imaging (mpMRI) features extracted from an auto-fixed volume of interest (VOI) that quantifies the phenotype of clinically significant (CS) peripheral zone (PZ) prostate cancer (PCa).

Methods

This study included 206 patients with 262 prospectively called mpMRI prostate imaging reporting and data system 3–5 PZ lesions. Gleason scores > 6 were defined as CS PCa. Features were extracted with an auto-fixed 12-mm spherical VOI placed around a pin point in each lesion. The value of dynamic contrast-enhanced imaging(DCE), multivariate feature selection and extreme gradient boosting (XGB) vs. univariate feature selection and random forest (RF), expert-based feature pre-selection, and the addition of image filters was investigated using the training (171 lesions) and test (91 lesions) datasets.

Results

The best model with features from T2-weighted (T2-w) + diffusion-weighted imaging (DWI) + DCE had an area under the curve (AUC) of 0.870 (95% CI 0.980–0.754). Removal of DCE features decreased AUC to 0.816 (95% CI 0.920–0.710), although not significantly (p = 0.119). Multivariate and XGB outperformed univariate and RF (p = 0.028). Expert-based feature pre-selection and image filters had no significant contribution.

Conclusions

The phenotype of CS PZ PCa lesions can be quantified using a radiomics approach based on features extracted from T2-w + DWI using an auto-fixed VOI. Although DCE features improve diagnostic performance, this is not statistically significant. Multivariate feature selection and XGB should be preferred over univariate feature selection and RF. The developed model may be a valuable addition to traditional visual assessment in diagnosing CS PZ PCa.

Key Points

• T2-weighted and diffusion-weighted imaging features are essential components of a radiomics model for clinically significant prostate cancer; addition of dynamic contrast-enhanced imaging does not significantly improve diagnostic performance.

• Multivariate feature selection and extreme gradient outperform univariate feature selection and random forest.

• The developed radiomics model that extracts multiparametric MRI features with an auto-fixed volume of interest may be a valuable addition to visual assessment in diagnosing clinically significant prostate cancer.

Electronic supplementary material

The online version of this article (10.1007/s00330-019-06488-y) contains supplementary material, which is available to authorized users.

Discovery of stable and prognostic CT-based radiomic features independent of contrast administration and dimensionality in oesophageal cancer

The aim of this work was to investigate radiomic analysis of contrast and non-contrast enhanced planning CT images of oesophageal cancer (OC) patients in terms of stability, dimensionality and contrast agent dependency. The prognostic significance of CT-based radiomic features was also evaluated. Different 2D and 3D radiomic features were extracted from contrast and non-contrast enhanced CT images of 213 patients from the multi-centre SCOPE1 randomised controlled trial (RCT) in OC. Feature stability was evaluated by randomly dividing patients into three groups and identifying textures with similar distributions among groups with a Kruskal-Wallis analysis. A paired two-sided Wilcoxon signed rank test was used to assess for significant differences in the remaining corresponding 2D and 3D stable features. A prognostic model was constructed using clinical characteristics and remaining filtered features. The discriminative ability of significant variables was tested using Kaplan-Meier analysis. A total of 238 2D and 3D radiomic features were computed from oesophageal CT images. More than 75 features were stable if extracted from homogeneous cohort (contrast or non-contrast enhanced CT images) and inhomogeneous cohort (contrast and non-contrast enhanced CT images). Among the remaining corresponding stable features computed from both cohorts, only 4 features did not show a statistically significant difference if obtained in 2D or in 3D (p-value < 0.05). A Cox regression model constructed using 5 clinical variables (age, sex, tumour, node and metastasis (TNM) stage, WHO performance status and contrast administration) and 4 radiomic variables (inverse varianceGLCM, large distance emphasisGLDZM, zone distance non uniformity normGLDZM, zone distance varianceGLDZM), identified one radiomic feature (zone distance varianceGLDZM) that was significantly associated with overall survival (p-value = 0.032, HR = 1.25, 95% CI = 1.02-1.52). A significant difference in overall survival between groups was found when considering a threshold of zone distance varianceGLDZM equals to 1.70 (X2 = 7.692, df = 1, p-value = 0.006). Zone distance varianceGLDZM was identified as the only stable CT radiomic feature statistically correlated with overall survival, independent of dimensionality and contrast administration. This feature was able to identify high-risk patients and if validated, could be the subject of a future clinical trial aiming to improve clinical decision making and personalise OC treatment.

Development of a predictive radiomics model for lymph node metastases in pre-surgical CT-based stage IA non-small cell lung cancer

OBJECTIVES

To develop and validate predictive models using clinical parameters, radiomic features and a combination of both for lymph node metastasis (LNM) in pre-surgical CT-based stage IA non-small cell lung cancer (NSCLC) patients.

METHODS

This retrospective study included 649 pre-surgical CT-based stage IA NSCLC patients from our hospital. One hundred and thirty-eight (21 %) of the 649 patients had LNM after surgery. A total of 396 radiomic features were extracted from the venous phase contrast enhanced computed tomography (CECT). The training group included 455 patients (97 with and 358 without LNM) and the testing group included 194 patients (41 with and 153 without LNM). The least absolute shrinkage and selection operator (LASSO) algorithm was used for radiomic feature selection. The random forest (RF) was used for model development. Three models (a clinical model, a radiomics model, and a combined model) were developed to predict LNM in early stage NSCLC patients. The area under the receiver operating characteristic (ROC) curve (AUC) value and decision curve analysis were used to evaluate the performance in LNM status (with or without LNM) using the three models.

RESULTS

The ROC analysis (also decision curve analysis) showed predictive performance for LNM of the radiomics model (AUC values for training and testing, respectively 0.898 and 0.851) and of the combined model (0.911 and 0.860, respectively). Both performed better than the clinical model (0.739 and 0.614, respectively; delong test p-values both<0.001).

CONCLUSION

A radiomics model using the venous phase of CE-CT has potential for predicting LNM in pre-surgical CT-based stage IA NSCLC patients.

Application of radiomics signature captured from pretreatment thoracic CT to predict brain metastases in stage III/IV ALK-positive non-small cell lung cancer patients

Background

The purpose of this study is to develop a radiomics approach to predict brain metastasis (BM) for stage III/IV ALK-positive non-small cell lung cancer (NSCLC) patients.

Methods

Patients with ALK-positive III/IV NSCLC from 2014 to 2017 were enrolled retrospectively. Their pretreatment thoracic CT images were collected, and the gross tumor volume (GTV) was defined by two experienced radiation oncologists. An in-house feature extraction code-set was performed based on MATLAB 2015b (Mathworks, Natick, MA, USA) in patients' CT images to extract features. Patients were randomly divided into training set and test set (4:1) by using createDataPartition function in caret package. A test-retest in RIDER NSCLC dataset was performed to identify stable radiomics features. LASSO Cox regression and a leave-one-out cross-validation were conducted to identify optimal features for the logistic regression model to evaluate the predictive value of radiomics feature(s) for BM. Furthermore, extended validation for the radiomics feature(s) and Cox regression analyses which combined radiomics feature(s) and treatment elements were implemented to predict the risk of BM during follow-up.

Results

In total, 132 patients were included, among which 27 patients had pretreatment BM. The median follow-up time was 11.8 (range, 0.1-65.2) months. In the training set, one radiomics feature (W_GLCM_LH_Correlation) showed discrimination ability of BM (P value =0.014, AUC =0.687, 95% CI: 0.551-0.824, specificity =83.5%, sensitivity =57.1%). It also exhibited reposeful performance in the test set (AUC =0.642, 95% CI: 0.501-0.783, specificity =60.0%, sensitivity =83.3%). Those 105 patients without pretreatment BM were divided into stage III (n=57) and stage IV (n=48) groups. The radiomics feature (W_GLCM_LH_Correlation) had moderate performance to predict BM during/after treatment in separate groups (stage III: AUC =0.682, 95% CI: 0.537-0.826, specificity =64.4%, sensitivity =75.0%; stage IV: AUC =0.653, 95% CI: 0.503-0.804, specificity =70.4%, sensitivity =75.0%). Meanwhile, stage III patients could be divided into low risk and high risk groups for BM during surveillance according to Cox regression analysis (log-rank P value =0.021).

Conclusions

We identified one wavelet texture feature derived from pretreatment thoracic CT that presented potential in predicting BM in stage III/IV ALK-positive NSCLC patients. This could be beneficial to risk stratification for such patients. Further investigation is necessary to include expanded sample size investigation and external multicenter validation.

Predictive Power of a Radiomic Signature Based on 18F-FDG PET/CT Images for EGFR Mutational Status in NSCLC

Radiomics has become an area of interest for tumor characterization in ¹⁸F-Fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) imaging. The aim of the present study was to demonstrate how imaging phenotypes was connected to somatic mutations through an integrated analysis of 115 non-small cell lung cancer (NSCLC) patients with somatic mutation testings and engineered computed PET/CT image analytics. A total of 38 radiomic features quantifying tumor morphological, grayscale statistic, and texture features were extracted from the segmented entire-tumor region of interest (ROI) of the primary PET/CT images. The ensembles for boosting machine learning scheme were employed for classification, and the least absolute shrink age and selection operator (LASSO) method was used to select the most predictive radiomic features for the classifiers. A radiomic signature based on both PET and CT radiomic features outperformed individual radiomic features, the PET or CT radiomic signature, and the conventional PET parameters including the maximum standardized uptake value (SUVmax), SUVmean, SUVpeak, metabolic tumor volume (MTV), and total lesion glycolysis (TLG), in discriminating between mutant-type of epidermal growth factor receptor (EGFR) and wild-type of EGFR- cases with an AUC of 0.805, an accuracy of 80.798%, a sensitivity of 0.826 and a specificity of 0.783. Consistently, a combined radiomic signature with clinical factors exhibited a further improved performance in EGFR mutation differentiation in NSCLC. In conclusion, tumor imaging phenotypes that are driven by somatic mutations may be predicted by radiomics based on PET/CT images.

Radiomics Analysis of Dynamic Contrast-Enhanced Magnetic Resonance Imaging for the Prediction of Sentinel Lymph Node Metastasis in Breast Cancer

Purpose: To investigate whether a combination of radiomics and automatic machine learning applied to dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) of primary breast cancer can non-invasively predict axillary sentinel lymph node (SLN) metastasis.

Methods: 62 patients who received a DCE-MRI breast scan were enrolled. Tumor resection and sentinel lymph node (SLN) biopsy were performed within 1 week after the DCE-MRI examination. According to the time signal intensity curve, the volumes of interest (VOIs) were delineated on the whole tumor in the images with the strongest enhanced phase. Datasets were randomly divided into two sets including a training set (~80%) and a validation set (~20%). A total of 1,409 quantitative imaging features were extracted from each VOI. The select K best and least absolute shrinkage and selection operator (Lasso) were used to obtain the optimal features. Three classification models based on the logistic regression (LR), XGboost, and support vector machine (SVM) classifiers were constructed. Receiver Operating Curve (ROC) analysis was used to analyze the prediction performance of the models. Both feature selection and models construction were firstly performed in the training set, then were further tested in the validation set by the same thresholds.

Results: There is no significant difference between all clinical and pathological variables in breast cancer patients with and without SLN metastasis (P > 0.05), except histological grade (P = 0.03). Six features were obtained as optimal features for models construction. In the validation set, with respect to the accuracy and MSE, the SVM demonstrated the highest performance, with an accuracy, AUC, sensitivity (for positive SLN), specificity (for positive SLN) and Mean Squared Error (MSE) of 0.85, 0.83, 0.71, 1, 0.26, respectively.

Conclusions: We demonstrated the feasibility of combining artificial intelligence and radiomics from DCE-MRI of primary tumors to predict axillary SLN metastasis in breast cancer. This non-invasive approach could be very promising in application.

Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- and PET-imaged head and neck cancer patients

Radiomics studies require many patients in order to power them, thus patients are often combined from different institutions and using different imaging protocols. Various studies have shown that imaging protocols affect radiomics feature values. We examined whether using data from cohorts with controlled imaging protocols improved patient outcome models. We retrospectively reviewed 726 CT and 686 PET images from head and neck cancer patients, who were divided into training or independent testing cohorts. For each patient, radiomics features with different preprocessing were calculated and two clinical variables-HPV status and tumor volume-were also included. A Cox proportional hazards model was built on the training data by using bootstrapped Lasso regression to predict overall survival. The effect of controlled imaging protocols on model performance was evaluated by subsetting the original training and independent testing cohorts to include only patients whose images were obtained using the same imaging protocol and vendor. Tumor volume, HPV status, and two radiomics covariates were selected for the CT model, resulting in an AUC of 0.72. However, volume alone produced a higher AUC, whereas adding radiomics features reduced the AUC. HPV status and one radiomics feature were selected as covariates for the PET model, resulting in an AUC of 0.59, but neither covariate was significantly associated with survival. Limiting the training and independent testing to patients with the same imaging protocol reduced the AUC for CT patients to 0.55, and no covariates were selected for PET patients. Radiomics features were not consistently associated with survival in CT or PET images of head and neck patients, even within patients with the same imaging protocol.

Predicting pathologic response to neoadjuvant chemoradiation in resectable stage III non-small cell lung cancer patients using computed tomography radiomic features

OBJECTIVE

The use of a neoadjuvant chemoradiation followed by surgery in patients with stage IIIA NSCLC is controversial and the benefit of surgery is limited. There are currently no clinically validated biomarkers to select patients for such an approach. In this study we evaluate computed tomography (CT) derived intratumoral and peritumoral texture and nodule shape features in their ability to predict major pathological response (MPR). MPR being defined as ≤10% of residual viable tumor, assessed at the time of surgery.

MATERIAL AND METHODS

Ninety patients with stage III NSCLC treated with chemoradiation prior to surgical resection were selected. The patients were divided randomly into two equal sets, one for training and one for independent testing. The radiomic texture and shape features were extracted from within the nodule (intra) and from the parenchymal regions immediately surrounding the nodule (peritumoral). A univariate regression analysis was performed on the image and clinicopathologic variables and then included into a multivariable logistic regression (MLR) for binary outcome prediction of MPR. The radiomic signature risk-score was generated by using a multivariate Cox regression model and association of the signature with OS and DFS was also evaluated.

RESULTS

Thirteen stable and predictive intratumoral and peritumoral radiomic texture features were found to be predictive of MPR. The MLR classifier yielded an AUC of 0.90 ± 0.025 within the training set and a corresponding AUC = 0.86 in prediction of MPR within the test set. The radiomic signature was also significantly associated with OS (HR = 11.18, 95% CI = 3.17, 44.1; p-value = 0.008) and DFS (HR = 2.78, 95% CI = 1.11, 4.12; p-value = 0.0042) in the testing set.

CONCLUSION

Texture features extracted within and around the lung tumor on CT images appears to be associated with the likelihood of MPR, OS and DFS to chemoradiation.

Radiomics for the prediction of EGFR mutation subtypes in non-small cell lung cancer

PURPOSE

This retrospective study was designed to investigate the ability of radiomics to predict the mutation status of epidermal growth factor receptor (EGFR) subtypes (19Del and L858R) in patients with non-small cell lung cancer (NSCLC).

METHODS

In total, 312 patients with NSCLC were included, and 580 radiomic features were extracted from the computed tomography images of each patient. In the training set, univariate analysis was performed on the clinical and radiomic features; logistic regression models were established using a 5-fold cross validation strategy for the prediction of EGFR subtypes 19Del and L858R. Subsequently, the predictive ability of the joint models was evaluated using the test set.

RESULTS

The results revealed that the radiomic features specific for EGFR 19Del and L858R were Gabor's MTRVariance, Gabor's PTREntropy, and sphericity. Additionally, the respective areas under the receiver operating characteristic curves of the EGFR 19Del and L858R joint models were 0.7925 and 0.7750 for the test set.

CONCLUSIONS

Our study demonstrated the potential for radiomics to predict EGFR 19Del and L858R. Epidermal growth factor receptor 19Del and L858R exhibited distinct imaging phenotypes, which may help to guide the selection of more accurate and personalized treatment programs for patients with NSCLC.

Convolution kernel and iterative reconstruction affect the diagnostic performance of radiomics and deep learning in lung adenocarcinoma pathological subtypes

Background

The aim of this study was to investigate the influence of convolution kernel and iterative reconstruction on the diagnostic performance of radiomics and deep learning (DL) in lung adenocarcinomas.

Methods

A total of 183 patients with 215 lung adenocarcinomas were included in this study. All CT imaging data was reconstructed with three reconstruction algorithms (ASiR at 0%, 30%, 60% strength), each with two convolution kernels (bone and standard). A total of 171 nodules were selected as the training‐validation set, whereas 44 nodules were selected as the testing set. Logistic regression and a DL framework‐DenseNets were selected to tackle the task. Three logical experiments were implemented to fully explore the influence of the studied parameters on the diagnostic performance. The receiver operating characteristic curve (ROC) was used to evaluate the performance of constructed models.

Results

In Experiments A and B, no statistically significant results were found in the radiomic method, whereas two and six pairs were statistically significant (P < 0.05) in the DL method. In Experiment_C, significant differences in one and four models were found in the radiomics and DL methods, respectively. Moreover, models constructed with standard convolution kernel data outperformed that constructed with bone convolution kernel data in all studied ASiR levels in the DL method. In the DL method, B0 and S60 performed best in bone and standard convolution kernel, respectively.

Conclusion

The results demonstrated that DL was more susceptible to CT parameter variability than radiomics. Standard convolution kernel images seem to be more appropriate for imaging analysis. Further investigation with a larger sample size is needed.

Decoding tumor mutation burden and driver mutations in early stage lung adenocarcinoma using CT-based radiomics signature

Background

Tumor mutation burden (TMB) is an important determinant and biomarker for response of targeted therapy and prognosis in patients with lung cancer. The present study aimed to determine whether radiomics signature could non‐invasively predict the TMB status and driver mutations in patients with resectable early stage lung adenocarcinoma (LUAD).

Methods

A total of 61pulmonary nodules (PNs) from 51 patients post‐operatively diagnosed LUAD were enrolled for analysis. Two datasets were divided according to two‐thirds of patients from different commercial Comprehensive Genomic Profiling (CGP) panels: a training cohort including 41 PNs and a testing cohort including rest 20PNs. We sequenced all tumor specimens and paired blood cells using next generation sequencing (NGS), so as to detect TMB status and somatic mutations. We collected 718 quantitative 3D radiomics features extracted from segmented volumes of each PNs and 78 clinical and pathological features retrieved from medical records as well. Support vector machine methods were performed to establish the predictive model.

Results

We established an efficient fusion‐positive tumor prediction model that predicts TMB status and EGFR/TP53 mutations of early stage LUAD. The radiomics signature yielded a median AUC value of 0.606, 0.604, and 0.586 respectively. Combining radiomics with the clinical information can further improve the prediction performance, which the median AUC values are 0.671 for TMB, 0.697 and 0.656 for EGFR/TP53 respectively.

Conclusion

It is feasible and effective to facilitate TMB and somatic driver mutations prediction by using the radiomics signature and NGS data in early stage LUAD.

Clinical impact of variability on CT radiomics and suggestions for suitable feature selection: a focus on lung cancer

Background

Radiomics suffers from feature reproducibility. We studied the variability of radiomics features and the relationship of radiomics features with tumor size and shape to determine guidelines for optimal radiomics study.

Methods

We dealt with 260 lung nodules (180 for training, 80 for testing) limited to 2 cm or less. We quantified how voxel geometry (isotropic/anisotropic) and the number of histogram bins, factors commonly adjusted in multi-center studies, affect reproducibility. First, features showing high reproducibility between the original and isotropic transformed voxel settings were identified. Second, features showing high reproducibility in various binning settings were identified. Two hundred fifty-two features were computed and features with high intra-correlation coefficient were selected. Features that explained nodule status (benign/malignant) were retained using the least absolute shrinkage selector operator. Common features among different settings were identified, and the final features showing high reproducibility correlated with nodule status were identified. The identified features were used for the random forest classifier to validate the effectiveness of the features. The properties of the uncalculated feature were inspected to suggest a tentative guideline for radiomics studies.

Results

Nine features showing high reproducibility for both the original and isotropic voxel settings were selected and used to classify nodule status (AUC 0.659–0.697). Five features showing high reproducibility among different binning settings were selected and used in classification (AUC 0.729–0.748). Some texture features are likely to be successfully computed if a nodule was larger than 1000 mm³.

Conclusions

Features showing high reproducibility among different settings correlated with nodule status were identified.

Electronic supplementary material

The online version of this article (10.1186/s40644-019-0239-z) contains supplementary material, which is available to authorized users.

Evaluation of an AI-Powered Lung Nodule Algorithm for Detection and 3D Segmentation of Primary Lung Tumors

Automated detection and segmentation is a prerequisite for the deployment of image-based secondary analyses, especially for lung tumors. However, currently only applications for lung nodules ≤3 cm exist. Therefore, we tested the performance of a fully automated AI-based lung nodule algorithm for detection and 3D segmentation of primary lung tumors in the context of tumor staging using the CT component of FDG-PET/CT and including all T-categories (T1-T4). FDG-PET/CTs of 320 patients with histologically confirmed lung cancer performed between 01/2010 and 06/2016 were selected. First, the main primary lung tumor within each scan was manually segmented using the CT component of the PET/CTs as reference. Second, the CT series were transferred to a platform with AI-based algorithms trained on chest CTs for detection and segmentation of lung nodules. Detection and segmentation performance were analyzed. Factors influencing detection rates were explored with binominal logistic regression and radiomic analysis. We also processed 94 PET/CTs negative for pulmonary nodules to investigate frequency and reasons of false-positive findings. The ratio of detected tumors was best in the T1-category (90.4%) and decreased continuously: T2 (70.8%), T3 (29.4%), and T4 (8.8%). Tumor contact with the pleura was a strong predictor of misdetection. Segmentation performance was excellent for T1 tumors (r = 0.908, p < 0.001) and tumors without pleural contact (r = 0.971, p < 0.001). Volumes of larger tumors were systematically underestimated. There were 0.41 false-positive findings per exam. The algorithm tested facilitates a reliable detection and 3D segmentation of T1/T2 lung tumors on FDG-PET/CTs. The detection and segmentation of more advanced lung tumors is currently imprecise due to the conception of the algorithm for lung nodules <3 cm. Future efforts should therefore focus on this collective to facilitate segmentation of all tumor types and sizes to bridge the gap between CAD applications for screening and staging of lung cancer.

Attenuation and Morphologic Characteristics Distinguishing a Ground-Glass Nodule Measuring 5-10 mm in Diameter as Invasive Lung Adenocarcinoma on Thin-Slice CT

OBJECTIVE. The purpose of this study is to comprehensively investigate the role of multiple features seen on thin-section CT (TSCT) in the differential diagnosis of ground-glass nodules (GGNs) measuring 5-10 mm in diameter as invasive adenocarcinoma (IAC). MATERIALS AND METHODS. The TSCT features of 313 surgically diagnosed GGNs from 288 patients were retrospectively reviewed. A logistic regression model was applied, and the AUC values for the model and the size and attenuation of the lesions were compared using ROC curve analysis. RESULTS. A total of 247 lung adenocarcinomas in situ (AISs) and minimally invasive adenocarcinomas (MIAs) (hereafter referred to as the AIS-MIA group) and 66 invasive adenocarcinomas (IACs) were identified. Compared with the AIS-MIA group, the IAC groups were significantly larger in size and had higher attenuation values, a higher frequency of mixed GGNs (all p < 0.001), bubblelike appearance, spiculation, pleural indentation, different locations, and a lower frequency of clear tumor-lung interface (all p < 0.05). The logistic model included size and attenuation (both p < 0.001; odds ratio [OR], 1.872 and 1.009, respectively) as well as tumor-lung interface (p = 0.001; OR, 0.242), bubblelike appearance (p < 0.05; OR, 2.205), and type of nodule. The AUC value for the logistic model was 0.847 (sensitivity, 80.3%; specificity, 81.0%) and was significantly higher than that for size or attenuation (both p < 0.01). CONCLUSION. Radiologic features could help in the differential diagnosis of a GGN that was 5-10 mm in diameter as IAC versus AIS or MIA. GGNs larger than 8.12 mm and with attenuation greater than -449.52 HU were more likely to be IAC.

Prognostic Impact of Longitudinal Monitoring of Radiomic Features in Patients with Advanced Non-Small Cell Lung Cancer

Tumor growth dynamics vary substantially in non-small cell lung cancer (NSCLC). We aimed to develop biomarkers reflecting longitudinal change of radiomic features in NSCLC and evaluate their prognostic power. Fifty-three patients with advanced NSCLC were included. Three primary variables reflecting patterns of longitudinal change were extracted: area under the curve of longitudinal change (AUC1), beta value reflecting slope over time, and AUC2, a value obtained by considering the slope and area over the longitudinal change of features. We constructed models for predicting survival with multivariate cox regression, and identified the performance of these models. AUC2 exhibited an excellent correlation between patterns of longitudinal volume change and a significant difference in overall survival time. Multivariate regression analysis based on cut-off values of radiomic features extracted from baseline CT and AUC2 showed that kurtosis of positive pixel values and surface area from baseline CT, AUC2 of density, skewness of positive pixel values, and entropy at inner portion were associated with overall survival. For the prediction model, the areas under the receiver operating characteristic curve (AUROC) were 0.948 and 0.862 at 1 and 3 years of follow-up, respectively. Longitudinal change of radiomic tumor features may serve as prognostic biomarkers in patients with advanced NSCLC.

Can texture features improve the differentiation of infiltrative lung adenocarcinoma appearing as ground glass nodules in contrast-enhanced CT?

OBJECTIVES

To investigate the validity and efficacy of comparing texture features from contrast-enhanced images with non-enhanced images in identifying infiltrative lung adenocarcinoma represented as ground glass nodules (GGN).

MATERIALS AND METHODS

A retrospective cohort study was conducted with patients presenting with lung adenocarcinoma and treated at a single centre between January 2015 to December 2017. All patients underwent standard and contrast-enhanced thoracic CT scans with 0.5 mm collimation and 1 mm slice reconstruction thickness before surgery. A total of 34 lung adenocarcinoma patients (representing 34 lesions) were analysed; including 21 instances of invasive adenocarcinoma (IAC) lesions, 4 instances of adenocarcinoma in situ (AIS) lesions, and 9 minimally invasive adenocarcinoma (MIA) lesions. After radiologists manually segmented the lesions, texture features were quantitatively extracted using Artificial Intelligence Kit (AK) software. Then, multivariate logistic regression analysis based on standard and contrast-enhanced CT texture features was employed to analyse the invasiveness of lung adenocarcinoma lesions appearing as GGNs. A receiver operating characteristic (ROC) curve analysis was used to evaluate the performance of those models.

RESULTS

A total of 21 quantitative texture features were extracted using the AK software. After dimensionality reduction, 5 and 3 features extracted from thin-section unenhanced and contrast-enhanced CT, respectively, were used to establish the model. The area under the ROC curve (AUC) values for unenhanced CT and enhanced CT features were 0.890 and 0.868, respectively. There was no significant difference (P = 0.190) in the AUC between models based on non-enhanced and contrast-enhanced CT texture features.

CONCLUSION

Compared with unenhanced CT, texture features extracted from contrast-enhanced CT provided no benefit in improving the differential diagnosis of infiltrative lung adenocarcinoma from non-infiltrative malignancies appearing as GGNs.

Radiomics and radiogenomics of prostate cancer

Radiomics and radiogenomics are attractive research topics in prostate cancer. Radiomics mainly focuses on extraction of quantitative information from medical imaging, whereas radiogenomics aims to correlate these imaging features to genomic data. The purpose of this review is to provide a brief overview summarizing recent progress in the application of radiomics-based approaches in prostate cancer and to discuss the potential role of radiogenomics in prostate cancer.

Biobanking in health care: evolution and future directions

BACKGROUND

The aim of the present review is to discuss how the promising field of biobanking can support health care research strategies. As the concept has evolved over time, biobanks have grown from simple biological sample repositories to complex and dynamic units belonging to large infrastructure networks, such as the Pan-European Biobanking and Biomolecular Resources Research Infrastructure (BBMRI). Biobanks were established to support scientific knowledge. Different professional figures with varied expertise collaborate to obtain and collect biological and clinical data from human subjects. At same time biobanks preserve the human and legal rights of each person that offers biomaterial for research.

METHODS

A literature review was conducted in April 2019 from the online database PubMed, accessed through the Bibliosan platform. Four primary topics related to biobanking will be discussed: (i) evolution, (ii) bioethical issues, (iii) organization, and (iv) imaging.

RESULTS

Most biobanks were founded as local units to support specific research projects, so they evolved in a decentralized manner. The consequence is an urgent needing for procedure harmonization regarding sample collection, processing, and storage. Considering the involvement of biomaterials obtained from human beings, different ethical issues such as the informed consent model, sample ownership, veto rights, and biobank sustainability are debated. In the face of these methodological and ethical challenges, international organizations such as BBMRI play a key role in supporting biobanking activities. Finally, a unique development is the creation of imaging biobanks that support the translation of imaging biomarkers (identified using a radiomic approach) into clinical practice by ensuring standardization of data acquisition and analysis, accredited technical validation, and transparent sharing of biological and clinical data.

CONCLUSION

Modern biobanks permit large-scale analysis for individuation of specific diseases biomarkers starting from biological or digital material (i.e., bioimages) with well-annotated clinical and biological data. These features are essential for improving personalized medical approaches, where effective biomarker identification is a critical step for disease diagnosis and prognosis.

Prediction of pathologic stage in non-small cell lung cancer using machine learning algorithm based on CT image feature analysis

PURPOSE

To explore imaging biomarkers that can be used for diagnosis and prediction of pathologic stage in non-small cell lung cancer (NSCLC) using multiple machine learning algorithms based on CT image feature analysis.

METHODS

Patients with stage IA to IV NSCLC were included, and the whole dataset was divided into training and testing sets and an external validation set. To tackle imbalanced datasets in NSCLC, we generated a new dataset and achieved equilibrium of class distribution by using SMOTE algorithm. The datasets were randomly split up into a training/testing set. We calculated the importance value of CT image features by means of mean decrease gini impurity generated by random forest algorithm and selected optimal features according to feature importance (mean decrease gini impurity > 0.005). The performance of prediction model in training and testing sets were evaluated from the perspectives of classification accuracy, average precision (AP) score and precision-recall curve. The predictive accuracy of the model was externally validated using lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) samples from TCGA database.

RESULTS

The prediction model that incorporated nine image features exhibited a high classification accuracy, precision and recall scores in the training and testing sets. In the external validation, the predictive accuracy of the model in LUAD outperformed that in LUSC.

CONCLUSIONS

The pathologic stage of patients with NSCLC can be accurately predicted based on CT image features, especially for LUAD. Our findings extend the application of machine learning algorithms in CT image feature prediction for pathologic staging and identify potential imaging biomarkers that can be used for diagnosis of pathologic stage in NSCLC patients.

Assessment of a Radiomic Signature Developed in a General NSCLC Cohort for Predicting Overall Survival of ALK-Positive Patients With Different Treatment Types

BACKGROUND

The purpose of the study was to investigate the potential of a radiomic signature developed in a general non-small-cell lung cancer (NSCLC) cohort for predicting the overall survival of anaplastic lymphoma kinase (ALK)-positive (ALK⁺) patients with different treatment types.

MATERIALS AND METHODS

After test-retest in the Reference Image Database to Evaluate Therapy Response data set, 132 features (intraclass correlation coefficient > 0.9) were selected in the least absolute shrinkage and selection operator Cox regression model with a leave-one-out cross-validation. The NSCLC radiomics collection from The Cancer Imaging Archive was randomly divided into a training set (n = 254) and a validation set (n = 63) to develop a general radiomic signature for NSCLC. In our ALK⁺ set, 35 patients received targeted therapy and 19 patients received nontargeted therapy. The developed signature was tested later in this ALK⁺ set. Performance of the signature was evaluated with the concordance index (C-index) and stratification analysis.

RESULTS

The general signature had good performance (C-index > 0.6; log rank P < .05) in the NSCLC radiomics collection. It includes 5 features: Geom_va_ratio, W_GLCM_Std, W_GLCM_DV, W_GLCM_IM2, and W_his_mean. Its accuracy of predicting overall survival in the ALK⁺ set achieved 0.649 (95% confidence interval [CI], 0.640-0.658). Nonetheless, impaired performance was observed in the targeted therapy group (C-index = 0.573; 95% CI, 0.556-0.589) whereas significantly improved performance was observed in the nontargeted therapy group (C-index = 0.832; 95% CI, 0.832-0.852). Stratification analysis also showed that the general signature could only identify high- and low-risk patients in the nontargeted therapy group (log rank P = .00028).

CONCLUSION

This preliminary study suggests that the applicability of a general signature to ALK⁺ patients is limited. The general radiomic signature seems to be only applicable to ALK⁺ patients who had received nontargeted therapy, which indicates that developing special radiomics signatures for patients treated with tyrosine kinase inhibitors might be necessary.

Machine Learning and Deep Neural Networks in Thoracic and Cardiovascular Imaging

Advances in technology have always had the potential and opportunity to shape the practice of medicine, and in no medical specialty has technology been more rapidly embraced and adopted than radiology. Machine learning and deep neural networks promise to transform the practice of medicine, and, in particular, the practice of diagnostic radiology. These technologies are evolving at a rapid pace due to innovations in computational hardware and novel neural network architectures. Several cutting-edge postprocessing analysis applications are actively being developed in the fields of thoracic and cardiovascular imaging, including applications for lesion detection and characterization, lung parenchymal characterization, coronary artery assessment, cardiac volumetry and function, and anatomic localization. Cardiothoracic and cardiovascular imaging lies at the technological forefront of radiology due to a confluence of technical advances. Enhanced equipment has enabled computed tomography and magnetic resonance imaging scanners that can safely capture images that freeze the motion of the heart to exquisitely delineate fine anatomic structures. Computing hardware developments have enabled an explosion in computational capabilities and in data storage. Progress in software and fluid mechanical models is enabling complex 3D and 4D reconstructions to not only visualize and assess the dynamic motion of the heart, but also quantify its blood flow and hemodynamics. And now, innovations in machine learning, particularly in the form of deep neural networks, are enabling us to leverage the increasingly massive data repositories that are prevalent in the field. Here, we discuss developments in machine learning techniques and deep neural networks to highlight their likely role in future radiologic practice, both in and outside of image interpretation and analysis. We discuss the concepts of validation, generalizability, and clinical utility, as they pertain to this and other new technologies, and we reflect upon the opportunities and challenges of bringing these into daily use.

Feasibility of an ADC-based radiomics model for predicting pelvic lymph node metastases in patients with stage IB-IIA cervical squamous cell carcinoma

OBJECTIVES

To investigate the prediction value of a radiomics model based on apparent diffusion coefficient (ADC) maps for pelvic lymph node metastasis (PLNM) in patients with stage IB-IIA cervical squamous cell carcinoma (CSCC).

METHODS

A total of 153 stage IB-IIA CSCC patients who underwent preoperative MRI including DWI from January 2015 to October 2017 were retrospectively studied and divided into a training cohort ( n = 102) and a validation cohort ( n = 51). Radiomics features were extracted from the ADC maps. The one-way ANOVA method, Mann-Whitney U test and Pearson's correlation analysis were used for selecting radiomics features. Logistic regression analyses were used to develop the model. ROC analyses were used to evaluate the prediction performance of the model.

RESULTS

Clinical stage, tumor diameter, and MR-reported lymph node (LN) status were significantly associated with LN status ( p < 0.05 for both the training and validation cohorts). The radiomics model, which incorporated clinical stage, MR-reported LN status, and grey-level non-uniformity, showed good predictive performance in the training group (AUC 0.864; 95% CI, 0.782 - 0.924) and the validation group (AUC 0.870; 95% CI, 0.747 - 0.948). The performance of the radiomics model was significantly better than that of each predictive factor alone.

CONCLUSION

The presented radiomics model, a non-invasive preoperative prediction tool, has the potential to have more predictive efficacy than clinical and radiological factors for differentiating between metastatic and non-metastatic lymph nodes.

ADVANCES IN KNOWLEDGE

A radiomics model derived from the ADC maps of primary lesions demonstrated good performance for predicting PLNM in stage IB-IIA CSCC patients and may help to improve clinical decision-making.

CT-Based Radiomics Model for Predicting Brain Metastasis in Category T1 Lung Adenocarcinoma

OBJECTIVE. The purpose of this study is to develop and evaluate an unenhanced CT-based radiomics model to predict brain metastasis (BM) in patients with category T1 lung adenocarcinoma. MATERIALS AND METHODS. A total of 89 eligible patients with category T1 lung adenocarcinoma were enrolled and classified as patients with BM (n = 35) or patients without BM (n = 54). A total of 1160 quantitative radiomic features were extracted from unenhanced CT images of each patient. Three prediction models (the clinical model, the radiomics model, and a hybrid [clinical plus radiomics] model) were established. The ROC AUC value and 10-fold cross-validation were used to evaluate the prediction performance of the models. RESULTS. In terms of predictive performance, the mean AUC value was 0.759 (95% CI, 0.643-0.867; sensitivity, 82.9%; specificity, 57.4%) for the clinical model, 0.847 (95% CI, 0.739-0.915; sensitivity, 80.0%; specificity, 81.5%) for the radiomics model, and 0.871 (95% CI, 0.767-0.933; sensitivity = 82.9%, specificity = 83.3%) for the hybrid model. The hybrid and radiomics models (p = 0.0072 and 0.0492, respectively) performed significantly better than the clinical model. No significant difference was found between the radiomics model and the hybrid model (p = 0.1022). CONCLUSION. A CT-based radiomics model presented good predictive performance and great potential for predicting BM in patients with category T1 lung adenocarcinoma. As a promising adjuvant tool, it can be helpful for guiding BM screening and thus benefiting personalized surveillance for patients with lung cancer.

Perinodular and Intranodular Radiomic Features on Lung CT Images Distinguish Adenocarcinomas from Granulomas

Purpose To evaluate ability of radiomic (computer-extracted imaging) features to distinguish non-small cell lung cancer adenocarcinomas from granulomas at noncontrast CT. Materials and Methods For this retrospective study, screening or standard diagnostic noncontrast CT images were collected for 290 patients (mean age, 68 years; range, 18-92 years; 125 men [mean age, 67 years; range, 18-90 years] and 165 women [mean age, 68 years; range, 33-92 years]) from two institutions between 2007 and 2013. Histopathologic analysis was available for one nodule per patient. Corresponding nodule of interest was identified on axial CT images by a radiologist with manual annotation. Nodule shape, wavelet (Gabor), and texture-based (Haralick and Laws energy) features were extracted from intra- and perinodular regions. Features were pruned to train machine learning classifiers with 145 patients. In a test set of 145 patients, classifier results were compared against a convolutional neural network (CNN) and diagnostic readings of two radiologists. Results Support vector machine classifier with intranodular radiomic features achieved an area under the receiver operating characteristic curve (AUC) of 0.75 on the test set. Combining radiomics of intranodular with perinodular regions improved the AUC to 0.80. On the same test set, CNN resulted in an AUC of 0.76. Radiologist readers achieved AUCs of 0.61 and 0.60, respectively. Conclusion Radiomic features from intranodular and perinodular regions of nodules can distinguish non-small cell lung cancer adenocarcinomas from benign granulomas at noncontrast CT. © RSNA, 2018 Online supplemental material is available for this article. See also the editorial by Nishino in this issue.

Mass Effect Deformation Heterogeneity (MEDH) on Gadolinium-contrast T1-weighted MRI is associated with decreased survival in patients with right cerebral hemisphere Glioblastoma: A feasibility study

Subtle tissue deformations caused by mass-effect in Glioblastoma (GBM) are often not visually evident, and may cause neurological deficits, impacting survival. Radiomic features provide sub-visual quantitative measures to uncover disease characteristics. We present a new radiomic feature to capture mass effect-induced deformations in the brain on Gadolinium-contrast (Gd-C) T1w-MRI, and their impact on survival. Our rationale is that larger variations in deformation within functionally eloquent areas of the contralateral hemisphere are likely related to decreased survival. Displacements in the cortical and subcortical structures were measured by aligning the Gd-C T1w-MRI to a healthy atlas. The variance of deformation magnitudes was measured and defined as Mass Effect Deformation Heterogeneity (MEDH) within the brain structures. MEDH values were then correlated with overall-survival of 89 subjects on the discovery cohort, with tumors on the right (n = 41) and left (n = 48) cerebral hemispheres, and evaluated on a hold-out cohort (n = 49 subjects). On both cohorts, decreased survival time was found to be associated with increased MEDH in areas of language comprehension, social cognition, visual perception, emotion, somato-sensory, cognitive and motor-control functions, particularly in the memory areas in the left-hemisphere. Our results suggest that higher MEDH in functionally eloquent areas of the left-hemisphere due to GBM in the right-hemisphere may be associated with poor-survival.

Machine Learning Models for Multiparametric Glioma Grading With Quantitative Result Interpretations

Gliomas are the most common primary malignant brain tumors in adults. Accurate grading is crucial as therapeutic strategies are often disparate for different grades and may influence patient prognosis. This study aims to provide an automated glioma grading platform on the basis of machine learning models. In this paper, we investigate contributions of multi-parameters from multimodal data including imaging parameters or features from the Whole Slide images (WSI) and the proliferation marker Ki-67 for automated brain tumor grading. For each WSI, we extract both visual parameters such as morphology parameters and sub-visual parameters including first-order and second-order features. On the basis of machine learning models, our platform classifies gliomas into grades II, III, and IV. Furthermore, we quantitatively interpret and reveal the important parameters contributing to grading with the Local Interpretable Model-Agnostic Explanations (LIME) algorithm. The quantitative analysis and explanation may assist clinicians to better understand the disease and accordingly to choose optimal treatments for improving clinical outcomes. The performance of our grading model was evaluated with cross-validation, which randomly divided the patients into non-overlapping training and testing sets and repeatedly validated the model on the different testing sets. The primary results indicated that this modular platform approach achieved the highest grading accuracy of 0.90 ± 0.04 with support vector machine (SVM) algorithm, with grading accuracies of 0.91 ± 0.08, 0.90 ± 0.08, and 0.90 ± 0.07 for grade II, III, and IV gliomas, respectively.

Statistical CT reconstruction using region-aware texture preserving regularization learning from prior normal-dose CT image

In some clinical applications, prior normal-dose CT (NdCT) images are available, and the valuable textures and structure features in them may be used to promote follow-up low-dose CT (LdCT) reconstruction. This study aims to learn texture information from the NdCT images and leverage it for follow-up LdCT image reconstruction to preserve textures and structure features. Specifically, the proposed reconstruction method first learns the texture information from those patches with similar structures in NdCT image, and the similar patches can be clustered by searching context features efficiently from the surroundings of the current patch. Then it utilizes redundant texture information from the similar patches as a priori knowledge to describe specific regions in the LdCT image. The advanced region-aware texture preserving prior is termed as 'RATP'. The main advantage of the PATP prior is that it can properly learn the texture features from available NdCT images and adaptively characterize the region-specific structures in the LdCT image. The experiments using patient data were performed to evaluate the performance of the proposed method. The proposed RATP method demonstrated superior performance in LdCT imaging compared to the filtered back projection (FBP) and statistical iterative reconstruction (SIR) methods using Gaussian regularization, Huber regularization and the original texture preserving regularization.

Radiomic signature of 18F fluorodeoxyglucose PET/CT for prediction of gastric cancer survival and chemotherapeutic benefits

We aimed to evaluate whether radiomic feature-based fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) positron emission tomography (PET) imaging signatures allow prediction of gastric cancer (GC) survival and chemotherapy benefits.

Methods: A total of 214 GC patients (training (n = 132) or validation (n = 82) cohort) were subjected to radiomic feature extraction (80 features). Radiomic features of patients in the training cohort were subjected to a LASSO cox analysis to predict disease-free survival (DFS) and overall survival (OS) and were validated in the validation cohort. A radiomics nomogram with the radiomic signature incorporated was constructed to demonstrate the incremental value of the radiomic signature to the TNM staging system for individualized survival estimation, which was then assessed with respect to calibration, discrimination, and clinical usefulness. The performance was assessed with concordance index (C-index) and integrated Brier scores.

Results: Significant differences were found between the high- and low-radiomic score (Rad-score) patients in 5-year DFS and OS in training and validation cohorts. Multivariate analysis revealed that the Rad-score was an independent prognostic factor. Incorporating the Rad-score into the radiomics-based nomogram resulted in better performance (C-index: DFS, 0.800; OS, 0.786; in the training cohort) than TNM staging system and clinicopathologic nomogram. Further analysis revealed that patients with higher Rad-scores were prone to benefit from chemotherapy.

Conclusion: The newly developed radiomic signature was a powerful predictor of OS and DFS. Moreover, the radiomic signature could predict which patients could benefit from chemotherapy.

Identification of optimal mother wavelets in survival prediction of lung cancer patients using wavelet decomposition-based radiomic features

PURPOSE

To identify the optimal mother wavelets in survival prediction of lung cancer patients using wavelet decomposition-based (WDB) radiomic features in CT images.

MATERIALS AND METHODS

The CT images of patients with histologically confirmed nonsmall cell lung carcinomas (NSCLCs) in training (Dataset T; n = 162) and validation (Dataset V; n = 143) datasets were analyzed for this study. The optimal mother wavelets were identified based on the impacts of the WDB radiomic features on the patient survival times. Four hundred and thirty-two three-dimensional WDB radiomic features were calculated from regions of interest (ROI) of 162 tumor contours. A Coxnet algorithm was used to select a subset of radiomic features (signature) based on the prediction of survival times with a fivefold cross validation. The impacts of the radiomic features on the patients' survival times were assessed by using a multivariate Cox proportional hazard regression (MCPHR) model. The major contribution of this study was to identify optimal mother wavelets based on a maximization of a novel ranking index (RI) incorporating the Coxnet cross-validated partial log-likelihood and the summation of the P-values of the radiomic features in the MCPHR model on Dataset T. The prognostic performance of the optimal mother wavelets was validated based on the concordance index (CI) of the MCPHR models when applied to Dataset V. The proposed approach was tested by using 31 mother wavelets from 6 wavelet families that were available in a commercially available software (Matlab^® 2016b).

RESULTS

The optimal mother wavelets were Symlet 5 and Biorthogonal 2.6 at 128 requantization levels, which yielded RIs of 4.27 ± 0.29 (3 features) and 6.50 ± 0.50 (5 features), respectively. The CIs of the MCPHR models of Symlet 5 were 0.66 ± 0.03 (Dataset T) and 0.64 ± 0.00 (Dataset V), whereas those of Biorthogonal 2.6 were 0.68 ± 0.03 (Dataset T) and 0.62 ± 0.02 (Dataset V). The radiomic signatures included the GLRLM-based LHL gray level nonuniformity feature that demonstrated statistically significant differences in stratifying patients with better and worse prognoses in Datasets T and V.

CONCLUSION

This study has revealed the potential of Symlet and Biorthogonal mother wavelets in  the survival prediction of lung cancer patients by using WDB radiomic features in CT images.

Quantitative vessel tortuosity: A potential CT imaging biomarker for distinguishing lung granulomas from adenocarcinomas

Adenocarcinomas and active granulomas can both have a spiculated appearance on computed tomography (CT) and both are often fluorodeoxyglucose (FDG) avid on positron emission tomography (PET) scan, making them difficult to distinguish. Consequently, patients with benign granulomas are often subjected to invasive surgical biopsies or resections. In this study, quantitative vessel tortuosity (QVT), a novel CT imaging biomarker to distinguish between benign granulomas and adenocarcinomas on routine non-contrast lung CT scans is introduced. Our study comprised of CT scans of 290 patients from two different institutions, one cohort for training (N = 145) and the other (N = 145) for independent validation. In conjunction with a machine learning classifier, the top informative and stable QVT features yielded an area under receiver operating characteristic curve (ROC AUC) of 0.85 in the independent validation set. On the same cohort, the corresponding AUCs for two human experts including a radiologist and a pulmonologist were found to be 0.61 and 0.60, respectively. QVT features also outperformed well known shape and textural radiomic features which had a maximum AUC of 0.73 (p-value = 0.002), as well as features learned using a convolutional neural network AUC = 0.76 (p-value = 0.028). Our results suggest that QVT features could potentially serve as a non-invasive imaging biomarker to distinguish granulomas from adenocarcinomas on non-contrast CT scans.

Prediction of chemotherapy response in ovarian cancer patients using a new clustered quantitative image marker

This study aimed to investigate the feasibility of integrating image features computed from both spatial and frequency domain to better describe the tumor heterogeneity for precise prediction of tumor response to postsurgical chemotherapy in patients with advanced-stage ovarian cancer. A computer-aided scheme was applied to first compute 133 features from five categories namely, shape and density, fast Fourier transform, discrete cosine transform (DCT), wavelet, and gray level difference method. An optimal feature cluster was then determined by the scheme using the particle swarm optimization algorithm aiming to achieve an enhanced discrimination power that was unattainable with the single features. The scheme was tested using a balanced dataset (responders and non-responders defined using 6 month PFS) retrospectively collected from 120 ovarian cancer patients. By evaluating the performance of the individual features among the five categories, the DCT features achieved the highest predicting accuracy than the features in other groups. By comparison, a quantitative image marker generated from the optimal feature cluster yielded the area under ROC curve (AUC) of 0.86, while the top performing single feature only had an AUC of 0.74. Furthermore, it was observed that the features computed from the frequency domain were as important as those computed from the spatial domain. In conclusion, this study demonstrates the potential of our proposed new quantitative image marker fused with the features computed from both spatial and frequency domain for a reliable prediction of tumor response to postsurgical chemotherapy.

Impact of Computer-Aided CT and PET Analysis on Non-invasive T Staging in Patients with Lung Cancer and Atelectasis

PURPOSE

Tumor delineation within an atelectasis in lung cancer patients is not always accurate. When T staging is done by integrated 2-deoxy-2-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG)-positron emission tomography (PET)/X-ray computer tomography (CT), tumors of neuroendocrine differentiation and slowly growing tumors can present with reduced FDG uptake, thus aggravating an exact T staging. In order to further exhaust information derived from [¹⁸F]FDG-PET/CT, we evaluated the impact of CT density and maximum standardized uptake value (SUVmax) for the classification of different tumor subtypes within a surrounding atelectasis, as well as possible cutoff values for the differentiation between the primary tumor and atelectatic lung tissue.

PROCEDURES

Seventy-two patients with histologically proven lung cancer and adjacent atelectasis were investigated. Non-contrast-enhanced [¹⁸F]FDG-PET/CT was performed within 2 weeks before surgery/biopsy. Boundaries of the primary within the atelectasis were determined visually on the basis of [¹⁸F]FDG uptake; CT density was quantified manually within each primary and each atelectasis.

RESULTS

CT density of the primary (36.4 Hounsfield units (HU) ± 6.2) was significantly higher compared to that of atelectatic lung (24.3 HU ± 8.3; p < 0.01), irrespective of the histological subtype. The discrimination between different malignant tumors using density analysis failed. SUVmax was increased in squamous cell carcinomas compared to adenocarcinomas. Irrespective of the malignant subtype, a possible cutoff value of 24 HU may help to exclude the presence of a primary in lesions below 24 HU, whereas a density above a threshold of 40 HU can help to exclude atelectatic lung.

CONCLUSION

Density measurements in patients with lung cancer and surrounding atelectasis may help to delineate the primary tumor, irrespective of the specific lung cancer subtype. This could improve T staging and radiation treatment planning (RTP) without additional application of a contrast agent in CT, or an additional magnetic resonance imaging (MRI), even in cases of lung tumors of neuroendocrine differentiation or in slowly growing tumors with less avidity to [¹⁸F]FDG.