Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol. 2017;14:749-762. [PMID: 28975929 DOI: 10.1038/nrclinonc.2017.141]

Pre-treatment radiomic features predict individual lymph node failure for head and neck cancer patients

BACKGROUND AND PURPOSE

To develop and validate a pre-treatment radiomics-based prediction model to identify pathological lymph nodes (pLNs) at risk of failures after definitive radiotherapy in head and neck squamous cell carcinoma patients.

MATERIALS AND METHODS

Training and validation cohorts consisted of 165 patients with 558 pLNs and 112 patients with 467 pLNs, respectively. All patients were primarily treated with definitive radiotherapy, with or without systemic treatment. The endpoint was the cumulative incidence of nodal failure. For each pLN, 82 pre-treatment CT radiomic features and 7 clinical features were included in the Cox proportional-hazard analysis.

RESULTS

There were 68 and 23 nodal failures in the training and validation cohorts, respectively. Multivariable analysis revealed three clinical features (T-stage, gender and WHO Performance-status) and two radiomic features (Least-axis-length representing nodal size and gray level co-occurrence matrix based - Correlation representing nodal heterogeneity) as independent prognostic factors. The model showed good discrimination with a c-index of 0.80 (0.69-0.91) in the validation cohort, significantly better than models based on clinical features (p < 0.001) or radiomics (p = 0.003) alone. High- and low-risk groups were defined by using thresholds of estimated nodal failure risks at 2-year of 60% and 10%, resulting in positive and negative predictive values of 94.4% and 98.7%, respectively.

CONCLUSION

A pre-treatment prediction model was developed and validated, integrating the quantitative radiomic features of individual lymph nodes with generally used clinical features. Using this prediction model, lymph nodes with a high failure risk can be identified prior to treatment, which might be used to select patients for intensified treatment strategies targeted on individual lymph nodes.

How to develop a meaningful radiomic signature for clinical use in oncologic patients

During the last decade, there is an increasing usage of quantitative methods in Radiology in an effort to reduce the diagnostic variability associated with a subjective manner of radiological interpretation. Combined approaches where visual assessment made by the radiologist is augmented by quantitative imaging biomarkers are gaining attention. Advances in machine learning resulted in the rise of radiomics that is a new methodology referring to the extraction of quantitative information from medical images. Radiomics are based on the development of computational models, referred to as “Radiomic Signatures”, trying to address either unmet clinical needs, mostly in the field of oncologic imaging, or to compare radiomics performance with that of radiologists. However, to explore this new technology, initial publications did not consider best practices in the field of machine learning resulting in publications with questionable clinical value. In this paper, our effort was concentrated on how to avoid methodological mistakes and consider critical issues in the workflow of the development of clinically meaningful radiomic signatures.

[Current Applications and Future Perspectives of Brain Tumor Imaging]

뇌종양의 진단 및 치료 반응 평가의 기본이 되는 영상기법은 해부학적 영상이다. 현재 임상에서 사용 가능한 영상기법들 중 확산 강조 영상 및 관류 영상이 추가적인 정보를 제공하고 있다. 최근에는 종양의 유전체 변이와 이질성 평가가 중요해지면서 라디오믹스와 딥러닝을 이용한 영상분석기법의 임상 응용이 기대되고 있다. 본 종설에서는 뇌종양 영상 임상 적용에서 여전히 중요한 해부학적 영상을 중심으로 한 자기공명영상 촬영 권고안, 최신 영상기법 중 확산 강조 영상 및 관류 영상의 기본 원리, 병태생리학적 배경 및 임상응용, 마지막으로 최근 컴퓨터 기술의 발전으로 많이 연구되고 있는 라디오믹스와 딥러닝의 뇌종양에서의 향후 활용가치에 대해 기술하고자 한다.

Quantitative Imaging Informatics for Cancer Research

PURPOSE

We summarize Quantitative Imaging Informatics for Cancer Research (QIICR; U24 CA180918), one of the first projects funded by the National Cancer Institute (NCI) Informatics Technology for Cancer Research program.

METHODS

QIICR was motivated by the 3 use cases from the NCI Quantitative Imaging Network. 3D Slicer was selected as the platform for implementation of open-source quantitative imaging (QI) tools. Digital Imaging and Communications in Medicine (DICOM) was chosen for standardization of QI analysis outputs. Support of improved integration with community repositories focused on The Cancer Imaging Archive (TCIA). Priorities included improved capabilities of the standard, toolkits and tools, reference datasets, collaborations, and training and outreach.

RESULTS

Fourteen new tools to support head and neck cancer, glioblastoma, and prostate cancer QI research were introduced and downloaded over 100,000 times. DICOM was amended, with over 40 correction proposals addressing QI needs. Reference implementations of the standard in a popular toolkit and standalone tools were introduced. Eight datasets exemplifying the application of the standard and tools were contributed. An open demonstration/connectathon was organized, attracting the participation of academic groups and commercial vendors. Integration of tools with TCIA was improved by implementing programmatic communication interface and by refining best practices for QI analysis results curation.

CONCLUSION

Tools, capabilities of the DICOM standard, and datasets we introduced found adoption and utility within the cancer imaging community. A collaborative approach is critical to addressing challenges in imaging informatics at the national and international levels. Numerous challenges remain in establishing and maintaining the infrastructure of analysis tools and standardized datasets for the imaging community. Ideas and technology developed by the QIICR project are contributing to the NCI Imaging Data Commons currently being developed.

Preoperative assessment of lymph node metastasis in Colon Cancer patients using machine learning: a pilot study

BACKGROUND

Preoperative detection of lymph node (LN) metastasis is critical for planning treatments in colon cancer (CC). The clinical diagnostic criteria based on the size of the LNs are not sensitive to determine metastasis using CT images. In this retrospective study, we investigated the potential value of CT texture features to diagnose LN metastasis using preoperative CT data and patient characteristics by developing quantitative prediction models.

METHODS

A total of 390 CC patients, undergone surgical resection, were enrolled in this monocentric study. 390 histologically validated LNs were collected from patients and randomly separated into training (312 patients, 155 metastatic and 157 normal LNs) and test cohorts (78 patients, 39 metastatic and 39 normal LNs). Six patient characteristics and 146 quantitative CT imaging features were analyzed and key variables were determined using either exhaustive search or least absolute shrinkage algorithm. Two kernel-based support vector machine classifiers (patient-characteristic model and radiomic-derived model), generated with 10-fold cross-validation, were compared with the clinical model that utilizes long-axis diameter for diagnosis of metastatic LN. The performance of the models was evaluated on the test cohort by computing accuracy, sensitivity, specificity, and area under the receiver operating curve (AUC).

RESULTS

The clinical model had an overall diagnostic accuracy of 64.87%; specifically, accuracy of 65.38% and 62.82%, sensitivity of 83.87% and 84.62%, and specificity of 47.13% and 41.03% for training and test cohorts, respectively. The patient-demographic model obtained accuracy of 67.31% and 73.08%, the sensitivity of 62.58% and 69.23%, and specificity of 71.97% and 76.23% for training and test cohorts, respectively. Besides, the radiomic-derived model resulted in an accuracy of 81.09% and 79.49%, sensitivity of 83.87% and 74.36%, and specificity of 78.34% and 84.62% for training and test cohorts, respectively. Furthermore, the diagnostic performance of the radiomic-derived model was significantly higher than clinical and patient-demographic models (p < 0.02) according to the DeLong method.

CONCLUSIONS

The texture of the LNs provided characteristic information about the histological status of the LNs. The radiomic-derived model leveraging LN texture provides better preoperative diagnostic accuracy for the detection of metastatic LNs compared to the clinically accepted diagnostic criteria and patient-demographic model.

Radiomics-Based Outcome Prediction for Pancreatic Cancer Following Stereotactic Body Radiotherapy

(1) Background: Radiomics use high-throughput mining of medical imaging data to extract unique information and predict tumor behavior. Currently available clinical prediction models poorly predict treatment outcomes in pancreatic adenocarcinoma. Therefore, we used radiomic features of primary pancreatic tumors to develop outcome prediction models and compared them to traditional clinical models. (2) Methods: We extracted and analyzed radiomic data from pre-radiation contrast-enhanced CTs of 74 pancreatic cancer patients undergoing stereotactic body radiotherapy. A panel of over 800 radiomic features was screened to create overall survival and local-regional recurrence prediction models, which were compared to clinical prediction models and models combining radiomic and clinical information. (3) Results: A 6-feature radiomic signature was identified that achieved better overall survival prediction performance than the clinical model (mean concordance index: 0.66 vs. 0.54 on resampled cross-validation test sets), and the combined model improved the performance slightly further to 0.68. Similarly, a 7-feature radiomic signature better predicted recurrence than the clinical model (mean AUC of 0.78 vs. 0.66). (4) Conclusion: Overall survival and recurrence can be better predicted with models based on radiomic features than with those based on clinical features for pancreatic cancer.

Elaboration of a multisequence MRI-based radiomics signature for the preoperative prediction of the muscle-invasive status of bladder cancer: a double-center study

OBJECTIVES

To develop a multisequence MRI-based radiomics signature for the preoperative prediction of the muscle-invasive status of bladder cancer (BCa).

METHODS

This retrospective study involved 106 eligible patients from two independent clinical centers. All patients underwent a preoperative 3.0 T MRI scan with T2-weighted image (T2WI) and multi-b-value diffusion-weighted image (DWI) sequences. In total, 1404 radiomics features were extracted from the largest region of the reported tumor locations on the T2WI, DWI, and corresponding apparent diffusion coefficient map (ADC) of each patient. A radiomics signature, namely the Radscore, was then generated using the recursive feature elimination approach and a logistic regression algorithm in a training cohort (n = 64). Its performance was then validated in an independent validation cohort (n = 42). The primary imaging and clinical factors in conjunction with the Radscore were used to determine whether the performance could be further improved.

RESULTS

The Radscore, generated by 36 selected radiomics features, demonstrated a favorable ability to predict muscle-invasive BCa status in both the training (AUC 0.880) and validation (AUC 0.813) cohorts. Subsequently, integrating the two independent predictors (including the Radscore and MRI-determined tumor stalk) into a nomogram exhibited more favorable discriminatory performance, with the AUC improved to 0.924 and 0.877 in both cohorts, respectively.

CONCLUSIONS

The proposed multisequence MRI-based radiomics signature alone could be an effective tool for quantitative prediction of muscle-invasive status of BCa. Integrating the Radscore with MRI-determined tumor stalk could further improve the discriminatory power, realizing more accurate prediction of nonmuscle-invasive and muscle-invasive BCa.

KEY POINTS

• DWI is superior to T2WI sequence in reflecting the heterogeneous differences between NMIBC and MIBC, and multisequence MRI helps in the preoperative prediction of muscle-invasive status of BCa. • Co-occurrence (CM), run-length matrix (RLM), and gray-level size zone matrix (GLSZM) features were the favorable feature categories for the prediction of muscle-invasive status of BCa. • The Radscore (proposed multisequence MRI-based radiomics signature) helps predict preoperatively muscle invasion. Combination with the MRI-determined tumor stalk further improves prediction.

Multiparametric MRI-Based Radiomics Nomogram for Predicting Lymphovascular Space Invasion in Endometrial Carcinoma

BACKGROUND

Lymphovascular space invasion (LVSI) of endometrial carcinoma (EMC) is one of the important prognostic factors, which is not usually visible subjectively. Therefore, clinicians lack imaging-based evidence about LVSI for preoperative treatment decision-making.

PURPOSE

To develop a multiparametric MRI (mpMRI)-based radiomics nomogram for predicting LVSI in EMC and provide decision-making support to clinicians.

STUDY TYPE

Retrospective.

POPULATION

In all, 144 patients with histologically confirmed EMC, 101 patients in a training cohort, and 43 patients in a test cohort.

FIELD STRENGTH/SEQUENCE

T₂ WI, contrast enhanced-T₁ WI, and diffusion-weighted imaging (DWI) at 3.0T MRI.

ASSESSMENT

Tumors were independently segmented images by two radiologists. Two pathologists reviewed the tissue specimens of the tumors to identify the existence of LVSI in consensus.

STATISTICAL TESTS

The intraclass correlation coefficient was used to test the reliability and least absolute shrinkage and selection operator (LASSO) regression for features selection and then developed a radiomics signature named Rad-score. A nomogram was developed in the training cohort. The diagnostic performance of the nomogram model was assessed by area under the curve (AUC) of the receiver operator characteristic (ROC) in the training and test cohort, respectively.

RESULTS

LVSI was identified in 32 patients (22.2%). Older age and high grade were correlated with LVSI at univariate analysis. There were five radiomics features that were identified as independent risk factors for LVSI by LASSO regression. Based on age, grade, and Rad-score, the AUC values of the nomogram model to predict LVSI in the training and test cohort were 0.820 (95% confidence interval [CI]: 0.725, 0.916; sensitivity: 82.6%, specificity: 72.9%), 0.807 (95% CI: 0.673, 0.941; sensitivity: 77.8%, specificity: 78.6%), respectively.

DATA CONCLUSION

The radiomic-based machine-learning model using a nomogram algorithm achieved high diagnostic performance for predicting LVSI of EMC preoperatively, which could enhance risk stratification and provide support for therapeutic decision-making.

LEVEL OF EVIDENCE

2.

TECHNICAL EFFICACY STAGE

3. J. Magn. Reson. Imaging 2020;52:1257-1262.

Integrated CT imaging and tissue immune features disclose a radio-immune signature with high prognostic impact on surgically resected NSCLC

OBJECTIVES

Qualitative and quantitative CT imaging features might intercept the multifaceted tumor immune microenvironment (TIME), providing a non-invasive approach to design new prognostic models in NSCLC patients.

MATERIALS AND METHODS

Our study population consisted of 100 surgically resected NSCLC patients among which 31 served as a validation cohort for quantitative image analysis. TIME was classified according to PD-L1 expression and the magnitude of Tumor Infiltrating Lymphocytes (TILs) and further defined as hot or cold by the tissue analysis of effector (CD8-to-CD3^high/PD-1-to-CD8^low) or inert (CD8-to-CD3^low/PD-1-to-CD8^high) phenotypes. CT datasets acted as source for qualitative (semantic, CT-SFs) and quantitative (radiomic, CT-RFs) features which were correlated with clinico-pathological and TIME profiles to determine their impact on survival outcome.

RESULTS

Specific CT-SFs (texture [TXT], effect [EFC] and margins [MRG]) strongly correlated to PD-L1 and TILs status and showed significant impact on survival outcome (TXT, HR:3.39, 95 % CI 1.12-10-27, P < 0.05; EFC, HR:0.41, 95 % CI 0.18-0.93, P < 0.05; MRG, HR:1.93, 95 % CI 0.88-4.25, P = 0.09). Seven CT derived radiomic features were able to sharply discriminate cases with hot (inflamed) vs cold (desert) TIME, which also exhibited opposite OS (long vs short, HR:0.09, 95 % CI 0.04-0.23, P < 0.001) and DFS (long vs short, HR:0.31, 95 % CI 0.16-0.58, P < 0.001). Moreover, we identified 6 prognostic radiomic features among which ClusterProminence displayed the highest statistical significance (HR:0.13, 95 % CI 0.06-0.31, P < 0.001). These findings were independently validated in an additional cohort of NSCLC (HR:0.11, 95 % CI 0.03-0.40, P = 0.001). Finally, in our training cohort we developed a multiparametric prognostic model, interlacing TIME and clinico-pathological characteristics with CT-SFs (ROC curve AUC:0.83, 95 % CI 0.71-0.92, P < 0.001) or CT-RFs (AUC: 0.91, 95 % CI 0.83-0.99, P < 0.001), which appeared to outperform pTNM staging (AUC: 0.66, 95 % CI 0.51-0.80, P < 0.05) in the risk assessment of NSCLC.

CONCLUSION

Higher order CT extracted features associated with specific TIME profiles may reveal a radio-immune signature with prognostic impact on resected NSCLC.

[Radiomics models based on non-enhanced MRI can differentiate chondrosarcoma from enchondroma]

OBJECTIVE

To develop and validate radiomics models based on non-enhanced magnetic resonance (MR) imaging for differentiating chondrosarcoma from enchondroma.

METHODS

We retrospectively evaluated a total of 68 patients (including 27 with chondrosarcoma and 41 with enchondroma), who were randomly divided into training group (n=46) and validation group (n=22). Radiomics features were extracted from T₁WI and T₂WI-FS sequences of the whole tumor by two radiologists independently and selected by Low Variance, Univariate feature selection, and least absolute shrinkage and selection operator (LASSO). Radiomics models were constructed by multivariate logistic regression analysis based on the features from T₁WI and T₂WI-FS sequences. The receiver-operating characteristics (ROC) curve and intraclass correlation coefficient (ICC) analyses of the radiomics models and conventional MR imaging were performed to determine their diagnostic accuracy.

RESULTS

The ICC value for interreader agreement of the radiomics features ranged from 0.779 to 0.923, which indicated good agreement. Ten and 11 features were selected from the T₁WI and T₂WI-FS sequences to construct radiomics models, respectively. The areas under the curve (AUCs) of T₁WI and T₂WI-FS models were 0.990 and 0.925 in training group and 0.915 and 0.855 in the validation group, respectively, showing no significant differences between the two sequence-based models (P&gt;0.05). In all the cases, the AUCs of the two radiomics models based on T₁WI and T₂WI-FS sequences and conventional MR imaging were 0.955, 0.901 and 0.569, respectively, demonstrating a significantly higher diagnostic accuracy of the two sequence-based radiomics models than conventional MR imaging (P&lt;0.01).

CONCLUSIONS

The radiomics models based on T₁WI and T₂WI-FS non-enhanced MR imaging can be used for the differentiation of chondrosarcoma from enchondroma.

Radiomics nomogram for preoperative differentiation of lung tuberculoma from adenocarcinoma in solitary pulmonary solid nodule

PURPOSE

To investigate the preoperative differential diagnostic performance of a radiomics nomogram in tuberculous granuloma (TBG) and lung adenocarcinoma (LAC) appearing as solitary pulmonary solid nodules (SPSNs).

METHOD

We retrospectively recruited 426 patients with SPSNs from two centers and assigned them to training (n = 123), internal validation (n = 121), and external validation cohorts (n = 182). A model of deep learning (DL) was built for tumor segmentation from routine computed tomography (CT) images and extraction of 3D radiomics features. We used the least absolute shrinkage and selection operator (LASSO) logistic regression to build a radiomics signature. A clinical model was developed with clinical factors, including age, gender, and CT-based subjective findings (eg, lesion size, lesion location, lesion margin, lobulated sharp, and spiculation sign). We constructed individualized radiomics nomograms incorporating the radiomics signature and clinical factors to validate the diagnostic ability.

RESULTS

Three factors - radiomics signature, age, and spiculation sign - were found to be independent predictors and were used to build the radiomics nomogram, which showed better diagnostic accuracy than any single model (all net reclassification improvement p < 0.05). The area under curve yielded was 0.9660 (95% confidence interval [CI], 0.9390-0.9931), 0.9342 (95% CI, 0.8944-0.9739), and 0.9064 (95% CI, 0.8639-0.9490) for the training, internal validation, and external validation cohorts, respectively. Decision curve analysis (DCA) and stratification analysis showed the nomogram has potential for generalizability.

CONCLUSION

The radiomics nomogram we developed can preoperatively distinguish between LAC and TBG in patient with a SPSN.

Global updates in the treatment of gastric cancer: a systematic review. Part 2: perioperative management, multimodal therapies, new technologies, standardization of the surgical treatment and educational aspects

Gastric cancer is the fifth malignancy and the third cause of cancer death worldwide, according to the global cancer statistics presented in 2018. Its definition and staging have been revised in the eight edition of the AJCC/TNM classification, which took effect in 2018. Novel molecular classifications for GC have been recently established and the process of translating these classifications into clinical practice is ongoing. The cornerstone of GC treatment is surgical, in a context of multimodal therapy. Surgical treatment is being standardized, and is evolving according to new anatomical concepts and to the recent technological developments. This is leading to a massive improvement in the use of mini-invasive techniques. Mini-invasive techniques aim to be equivalent to open surgery from an oncologic point of view, with better short-term outcomes. The persecution of better short-term outcomes also includes the optimization of the perioperative management, which is being implemented on large scale according to the enhanced recovery after surgery principles. In the era of precision medicine, multimodal treatment is also evolving. The long-time-awaited results of many trials investigating the role for preoperative and postoperative management have been published, changing the clinical practice. Novel investigations focused both on traditional chemotherapeutic regimens and targeted therapies are currently ongoing. Modern platforms increase the possibility for further standardization of the different treatments, promote the use of big data and open new possibilities for surgical learning. This systematic review in two parts assesses all the current updates in GC treatment.

Identification of Pancreaticoduodenectomy Resection for Pancreatic Head Adenocarcinoma: A Preliminary Study of Radiomics

Background

In a pathological examination of pancreaticoduodenectomy for pancreatic head adenocarcinoma, a resection margin without cancer cells in 1 mm is recognized as R0; a resection margin with cancer cells in 1 mm is recognized as R1. The preoperative identification of R0 and R1 is of great significance for surgical decision and prognosis. We conducted a preliminary radiomics study based on preoperative CT (computer tomography) images to evaluate a resection margin which was R0 or R1.

Methods

We retrospectively analyzed 258 preoperative CT images of 86 patients (34 cases of R0 and 52 cases of R1) who were diagnosed as pancreatic head adenocarcinoma and underwent pancreaticoduodenectomy. The radiomics study consists of five stages: (i) delineate and segment regions of interest (ROIs); (ii) by solving discrete Laplacian equations with Dirichlet boundary conditions, fit the ROIs to rectangular regions; (iii) enhance the textures of the fitted ROIs combining wavelet transform and fractional differential; (iv) extract texture features from the enhanced ROIs combining wavelet transform and statistical analysis methods; and (v) reduce features using principal component analysis (PCA) and classify the resection margins using the support vector machine (SVM), and then investigate the associations between texture features and histopathological characteristics using the Mann-Whitney U-test. To reduce overfitting, the SVM classifier embedded a linear kernel and adopted the leave-one-out cross-validation.

Results

It achieved an AUC (area under receiver operating characteristic curve) of 0.8614 and an accuracy of 84.88%. Setting p ≤ 0.01 in the Mann-Whitney U-test, two features of the run-length matrix, which are derived from diagonal sub-bands in wavelet decomposition, showed statistically significant differences between R0 and R1.

Conclusions

It indicates that the radiomics study is rewarding for the aided diagnosis of R0 and R1. Texture features can potentially enhance physicians' diagnostic ability.

Correlation Between Mammographic Radiomics Features and the Level of Tumor-Infiltrating Lymphocytes in Patients With Triple-Negative Breast Cancer

Objectives: Tumor-infiltrating lymphocytes (TILs) have been identified as a significant prognostic indicator of response to neoadjuvant therapy and immunotherapy for triple-negative breast cancer (TNBC) patients. Herein, we aim to assess the association between TIL levels and mammographic features in TNBC patients. Methods: Forty-three patients with surgically proven TNBC who underwent preoperative mammography from January 2018 to December 2018 were recruited. Pyradiomics software was used to extract 204 quantitative radiomics features, including morphologic, grayscale, and textural features, from the segmented lesion areas. The correlation between radiological characteristics and TIL levels was evaluated by screening the most statistically significant radiological features using Mann-Whitney U-test and Pearson correlation coefficient. The patients were divided into two groups based on tumor TIL levels: patients with TIL levels <50% and those with TIL levels ≥50%. The correlation between TIL levels and clinicopathological characteristics was assessed using the chi-square test or Fisher's exact test. Mann-Whitney U-test and Pearson correlation coefficient were used to analyze the statistical significance and Pearson correlation coefficient of clinical pathological features, age, and radiological features. Results: Of 43 patients, 32 (74.4%) had low TIL levels and 11 (25.6%) had high TIL levels. The histological grade of the low TIL group was higher than that of the high TIL group (p = 0.043). The high TIL group had a more negative threshold Ki-67 level (<14%) than the low TIL group (p = 0.017). The six most important radiomics features [uniformity, variance, grayscale symbiosis matrix (GLCM) correlation, GLCM autocorrelation, gray level difference matrix (GLDM) low gray level emphasis, and neighborhood gray-tone difference matrix (NGTDM) contrast], representing qualitative mammographic image characteristics, were statistically different (p < 0.05) among the low and high TIL groups. Tumors in the high TIL group had a more non-uniform density and a smoother gradient of the tumor pattern than the low TIL group. The changes in Ki-67, age, epidermal growth factor receptor, radiomic characteristics, and Pearson correlation coefficient were statistically significant (p < 0.05). Conclusion: Mammography features not only distinguish high and low TIL levels in TNBC patients but also can act as imaging biomarkers to enhance diagnosis and the response of patients to neoadjuvant therapies and immunotherapies.

Deep learning radiomic nomogram can predict the number of lymph node metastasis in locally advanced gastric cancer: an international multicenter study

BACKGROUND

Preoperative evaluation of the number of lymph node metastasis (LNM) is the basis of individual treatment of locally advanced gastric cancer (LAGC). However, the routinely used preoperative determination method is not accurate enough.

PATIENTS AND METHODS

We enrolled 730 LAGC patients from five centers in China and one center in Italy, and divided them into one primary cohort, three external validation cohorts, and one international validation cohort. A deep learning radiomic nomogram (DLRN) was built based on the images from multiphase computed tomography (CT) for preoperatively determining the number of LNM in LAGC. We comprehensively tested the DLRN and compared it with three state-of-the-art methods. Moreover, we investigated the value of the DLRN in survival analysis.

RESULTS

The DLRN showed good discrimination of the number of LNM on all cohorts [overall C-indexes (95% confidence interval): 0.821 (0.785-0.858) in the primary cohort, 0.797 (0.771-0.823) in the external validation cohorts, and 0.822 (0.756-0.887) in the international validation cohort]. The nomogram performed significantly better than the routinely used clinical N stages, tumor size, and clinical model (P < 0.05). Besides, DLRN was significantly associated with the overall survival of LAGC patients (n = 271).

CONCLUSION

A deep learning-based radiomic nomogram had good predictive value for LNM in LAGC. In staging-oriented treatment of gastric cancer, this preoperative nomogram could provide baseline information for individual treatment of LAGC.

Making Qualitative Intestinal Stricture Quantitative: Embracing Radiomics in IBD

Quantifying intestinal fibrosis is an important but challenging issue in stricturing CD. With the advent of radiomics, MRE and CTE are highly minable data rather than merely pictures, which will greatly facilitate precision medicine in intestinal fibrosis of CD.

Radiomics, Tumor Volume, and Blood Biomarkers for Early Prediction of Pseudoprogression in Patients with Metastatic Melanoma Treated with Immune Checkpoint Inhibition

PURPOSE

We assessed the predictive potential of positron emission tomography (PET)/CT-based radiomics, lesion volume, and routine blood markers for early differentiation of pseudoprogression from true progression at 3 months.

EXPERIMENTAL DESIGN

112 patients with metastatic melanoma treated with immune checkpoint inhibition were included in our study. Median follow-up duration was 22 months. 716 metastases were segmented individually on CT and 2[18F]fluoro-2-deoxy-D-glucose (FDG)-PET imaging at three timepoints: baseline (TP0), 3 months (TP1), and 6 months (TP2). Response was defined on a lesion-individual level (RECIST 1.1) and retrospectively correlated with FDG-PET/CT radiomic features and the blood markers LDH/S100. Seven multivariate prediction model classes were generated.

RESULTS

Two-year (median) overall survival, progression-free survival, and immune progression-free survival were 69% (not reached), 24% (6 months), and 42% (16 months), respectively. At 3 months, 106 (16%) lesions had progressed, of which 30 (5%) were identified as pseudoprogression at 6 months. Patients with pseudoprogressive lesions and without true progressive lesions had a similar outcome to responding patients and a significantly better 2-year overall survival of 100% (30 months), compared with 15% (10 months) in patients with true progressions/without pseudoprogression (P = 0.002). Patients with mixed progressive/pseudoprogressive lesions were in between at 53% (25 months). The blood prediction model (LDH+S100) achieved an AUC = 0.71. Higher LDH/S100 values indicated a low chance of pseudoprogression. Volume-based models: AUC = 0.72 (TP1) and AUC = 0.80 (delta-volume between TP0/TP1). Radiomics models (including/excluding volume-related features): AUC = 0.79/0.78. Combined blood/volume model: AUC = 0.79. Combined blood/radiomics model (including volume-related features): AUC = 0.78. The combined blood/radiomics model (excluding volume-related features) performed best: AUC = 0.82.

CONCLUSIONS

Noninvasive PET/CT-based radiomics, especially in combination with blood parameters, are promising biomarkers for early differentiation of pseudoprogression, potentially avoiding added toxicity or delayed treatment switch.

Radiomics: A primer for the radiation oncologist

PURPOSE

Radiomics are a set of methods used to leverage medical imaging and extract quantitative features that can characterize a patient's phenotype. All modalities can be used with several different software packages. Specific informatics methods can then be used to create meaningful predictive models. In this review, we will explain the major steps of a radiomics analysis pipeline and then present the studies published in the context of radiation therapy.

METHODS

A literature review was performed on Medline using the search engine PubMed. The search strategy included the search terms "radiotherapy", "radiation oncology" and "radiomics". The search was conducted in July 2019 and reference lists of selected articles were hand searched for relevance to this review.

RESULTS

A typical radiomics workflow always includes five steps: imaging and segmenting, data curation and preparation, feature extraction, exploration and selection and finally modeling. In radiation oncology, radiomics studies have been published to explore different clinical outcome in lung (n=5), head and neck (n=5), esophageal (n=3), rectal (n=3), pancreatic (n=2) cancer and brain metastases (n=2). The quality of these retrospective studies is heterogeneous and their results have not been translated to the clinic.

CONCLUSION

Radiomics has a great potential to predict clinical outcome and better personalize treatment. But the field is still young and constantly evolving. Improvement in bias reduction techniques and multicenter studies will hopefully allow more robust and generalizable models.

Deep learning-based radiomic features for improving neoadjuvant chemoradiation response prediction in locally advanced rectal cancer

Radiomic features achieve promising results in cancer diagnosis, treatment response prediction, and survival prediction. Our goal is to compare the handcrafted (explicitly designed) and deep learning (DL)-based radiomic features extracted from pre-treatment diffusion-weighted magnetic resonance images (DWIs) for predicting neoadjuvant chemoradiation treatment (nCRT) response in patients with locally advanced rectal cancer (LARC). 43 Patients receiving nCRT were included. All patients underwent DWIs before nCRT and total mesorectal excision surgery 6-12 weeks after completion of nCRT. Gross tumor volume (GTV) contours were drawn by an experienced radiation oncologist on DWIs. The patient-cohort was split into the responder group (n = 22) and the non-responder group (n = 21) based on the post-nCRT response assessed by postoperative pathology, MRI or colonoscopy. Handcrafted and DL-based features were extracted from the apparent diffusion coefficient (ADC) map of the DWI using conventional computer-aided diagnosis methods and a pre-trained convolution neural network, respectively. Least absolute shrinkage and selection operator (LASSO)-logistic regression models were constructed using extracted features for predicting treatment response. The model performance was evaluated with repeated 20 times stratified 4-fold cross-validation using receiver operating characteristic (ROC) curves and compared using the corrected paired t-test. The model built with handcrafted features achieved the mean area under the ROC curve (AUC) of 0.64, while the one built with DL-based features yielded the mean AUC of 0.73. The corrected paired t-test on AUC showed P-value < 0.05. DL-based features extracted from pre-treatment DWIs achieved significantly better classification performance compared with handcrafted features for predicting nCRT response in patients with LARC.

Advances in the Diagnosis and Management of Well-Differentiated Neuroendocrine Neoplasms

Neuroendocrine neoplasms constitute a diverse group of tumors that derive from the sensory and secretory neuroendocrine cells and predominantly arise within the pulmonary and gastrointestinal tracts. The majority of these neoplasms have a well-differentiated grade and are termed neuroendocrine tumors (NETs). This subgroup is characterized by limited proliferation and patients affected by these tumors carry a good to moderate prognosis. A substantial subset of patients presenting with a NET suffer from the consequences of endocrine syndromes as a result of the excessive secretion of amines or peptide hormones, which can impair their quality of life and prognosis. Over the past 15 years, critical developments in tumor grading, diagnostic biomarkers, radionuclide imaging, randomized controlled drug trials, evidence-based guidelines, and superior prognostic outcomes have substantially altered the field of NET care. Here, we review the relevant advances to clinical practice that have significantly upgraded our approach to NET patients, both in diagnostic and in therapeutic options.

Radiomics based on artificial intelligence in liver diseases: where we are?

Radiomics uses computers to extract a large amount of information from different types of images, form various quantifiable features, and select relevant features using artificial-intelligence algorithms to build models, in order to predict the outcomes of clinical problems (such as diagnosis, treatment, prognosis, etc.). The study of liver diseases by radiomics will contribute to early diagnosis and treatment of liver diseases and improve survival and cure rates of liver diseases. This field is currently in the ascendant and may have great development in the future. Therefore, we summarize the progress of current research in this article and then point out the related deficiencies and the direction of future research.

Conventional magnetic resonance imaging-based radiomic signature predicts telomerase reverse transcriptase promoter mutation status in grade II and III gliomas

PURPOSE

Telomerase reverse transcriptase (TERT) promoter mutation status is an important biomarker for the precision diagnosis and prognosis prediction of lower grade glioma (LGG). This study aimed to construct a radiomic signature to noninvasively predict the TERT promoter status in LGGs.

METHODS

Eighty-three local patients with pathology-confirmed LGG were retrospectively included as a training cohort, and 33 patients from The Cancer Imaging Archive (TCIA) were used as for independent validation. Three types of regions of interest (ROIs), which covered the tumor, peri-tumoral area, and tumor plus peri-tumoral area, were delineated on three-dimensional contrast-enhanced T1 (3D-CE-T1)-weighted and T2-weighted images. One hundred seven shape, first-order, and texture radiomic features from each modality under each ROI were extracted and selected through least absolute shrinkage and selection operator. Radiomic signatures were constructed with multiple classifiers and evaluated using receiver operating characteristic (ROC) analysis. The tumors were also stratified according to IDH status.

RESULTS

Three radiomic signatures, namely, tumoral radiomic signature, tumoral plus peri-tumoral radiomic signature, and fusion radiomic signature, were built, all of which exhibited good accuracy and balanced sensitivity and specificity. The tumoral signature displayed the best performance, with area under the ROC curves (AUC) of 0.948 (0.903-0.993) in the training cohort and 0.827 (0.667-0.988) in the validation cohort. In the IDH subgroups, the AUCs of the tumoral signature ranged from 0.750 to 0.940.

CONCLUSION

The MRI-based radiomic signature is reliable for noninvasive evaluation of TERT promoter mutations in LGG regardless of the IDH status. The inclusion of peri-tumoral area did not significantly improve the performance.

Predicting the Outcome of Transcatheter Arterial Embolization Therapy for Unresectable Hepatocellular Carcinoma Based on Radiomics of Preoperative Multiparameter MRI

BACKGROUND

In unresectable hepatocellular carcinoma (HCC), methods to predict patients at increased risk of progression are required.

PURPOSE

To investigate the feasibility of radiomics model in predicting early progression of unresectable HCC after transcatheter arterial chemoembolization (TACE) therapy using preoperative multiparametric magnetic resonance imaging (MP-MRI).

STUDY TYPE

Retrospective.

POPULATION

A total of 84 patients with BCLC B stage HCC from one medical center. According to the modified response evaluation criteria in solid tumors, patients who progressed at 6 months after TACE therapy were assigned as the progressive disease (PD) group (n = 32). Patients whose MRI was performed on four devices were divided into a training cohort (n = 67). Patients whose MRI was performed on other than the previous four devices were used as the testing set (n = 17).

FIELD STRENGTH/SEQUENCE

3.0T, 1.5T axial T₂ -weighted imaging (T₂ WI), diffusion-weighted imaging (DWI, b = 0, 500 s/mm² ), and apparent diffusion coefficient (ADC) ASSESSMENT: PD was confirmed via imaging studies with MRI. Risk factors, including age, alpha fetoprotein (AFP), size, and radiomic-related features of PD were assessed. In addition, the discrimination ability of each radiomics signature was tested on an independent testing set.

STATISTICAL TESTS

The area under the receiver-operator characteristic (ROC) curve (AUC) was used to evaluate the predictive accuracy of the radiomic signature in both the training and testing sets. The results indicated that the MP-MRI model achieved the greatest benefit.

RESULTS

In the testing set, the model based on DWI features presented an AUC of (b = 0, 0.786; b = 500, 0.729), followed by T₂ WI features (0.729) and ADC (0.714). The AUC of the MP-MRI signature was increased to 0.800 compared to any single MRI signature. The multivariate logistic analysis identified the radiomics signature as independent parameters of PD, while clinical information such as age, AFP, size, etc., had no significance in the PD group.

DATA CONCLUSION

Preoperative MP-MRI has the potential to predict the outcome of TACE therapy for unresectable HCC. In addition, these image features may be complementary to the current staging systems of HCC patients.

LEVEL OF EVIDENCE

2.

TECHNICAL EFFICACY STAGE

3. J. Magn. Reson. Imaging 2020;52:1083-1090.

Noninvasive imaging evaluation of tumor immune microenvironment to predict outcomes in gastric cancer

BACKGROUND

The tumor immune microenvironment can provide prognostic and predictive information. A previously validated ImmunoScore of Gastric Cancer (IS_GC) evaluates both lymphoid and myeloid cells in the tumor core and invasive margin with immunohistochemical staining of surgical specimens. We aimed to develop a noninvasive radiomics-based predictor of IS_GC.

PATIENTS AND METHODS

In this retrospective study including four independent cohorts of 1778 patients, we extracted 584 quantitative features from the intratumoral and peritumoral regions on contrast-enhanced computed tomography images. A radiomic signature [radiomics ImmunoScore (RIS)] was constructed to predict IS_GC using regularized logistic regression. We further evaluated its association with prognosis and chemotherapy response.

RESULTS

A 13-feature radiomic signature for IS_GC was developed and validated in three independent cohorts (area under the curve = 0.786, 0.745, and 0.766). The RIS signature was significantly associated with both disease-free and overall survival in the training and all validation cohorts [hazard ratio (HR) range: 0.296-0.487, all P < 0.001]. In multivariable analysis, the RIS remained an independent prognostic factor adjusting for clinicopathologic variables (adjusted HR range: 0.339-0.605, all P < 0.003). For stage II and stage III disease, patients with a high RIS derived survival benefit from adjuvant chemotherapy {HR = 0.436 [95% confidence interval (CI) 0.253-0.753], P = 0.002; HR = 0.591 (95% CI 0.428-0.818), P < 0.001, respectively}, whereas those with a low RIS did not.

CONCLUSION

The RIS is a reliable tool for evaluation of immunoscore and retains the prognostic significance in gastric cancer. Future prospective studies are required to confirm its potential to predict treatment response and select patients who will benefit from chemotherapy.

External Validation of an MRI-Derived Radiomics Model to Predict Biochemical Recurrence after Surgery for High-Risk Prostate Cancer

Adjuvant radiotherapy after prostatectomy was recently challenged by early salvage radiotherapy, which highlighted the need for biomarkers to improve risk stratification. Therefore, we developed an MRI ADC map-derived radiomics model to predict biochemical recurrence (BCR) and BCR-free survival (bRFS) after surgery. Our goal in this work was to externally validate this radiomics-based prediction model.

EXPERIMENTAL DESIGN

A total of 195 patients with a high recurrence risk of prostate cancer (pT3-4 and/or R1 and/or Gleason's score > 7) were retrospectively included in two institutions. Patients with postoperative PSA (Prostate Specific Antigen) > 0.04 ng/mL or lymph node involvement were excluded. Radiomics features were extracted from T2 and ADC delineated tumors. A total of 107 patients from Institution 1 were used to retrain the previously published model. The retrained model was then applied to 88 patients from Institution 2 for external validation. BCR predictions were evaluated using AUC (Area Under the Curve), accuracy, and bRFS using Kaplan-Meier curves.

RESULTS

With a median follow-up of 46.3 months, 52/195 patients experienced BCR. In the retraining cohort, the clinical prediction model (combining the number of risk factors and postoperative PSA) demonstrated moderate predictive power (accuracy of 63%). The radiomics model (ADC-based SZEGLSZM) predicted BCR with an accuracy of 78% and allowed for significant stratification of patients for bRFS (p < 0.0001). In Institution 2, this radiomics model remained predictive of BCR (accuracy of 0.76%) contrary to the clinical model (accuracy of 0.56%).

CONCLUSIONS

The recently developed MRI ADC map-based radiomics model was validated in terms of its predictive accuracy of BCR and bRFS after prostatectomy in an external cohort.

Liver-specific 3D sectioning molds for correlating in vivo CT and MRI with tumor histopathology in woodchucks (Marmota monax)

Purpose

To evaluate the spatial registration and correlation of liver and tumor histopathology sections with corresponding in vivo CT and MRI using 3D, liver-specific cutting molds in a woodchuck (Marmota monax) hepatic tumor model.

Methods

Five woodchucks chronically infected with woodchuck hepatitis virus following inoculation at birth and with confirmed hepatic tumors were imaged by contrast enhanced CT or MRI. Virtual 3D liver or tumor models were generated by segmentation of in vivo CT or MR imaging. A specimen-specific cavity was created inside a block containing cutting slots aligned with an imaging plane using computer-aided design software, and the final cutting molds were fabricated using a 3D printer. Livers were resected two days after initial imaging, fixed with formalin or left unfixed, inserted into the 3D molds, and cut into parallel pieces by passing a sharp blade through the parallel slots in the mold. Histopathology sections were acquired and their spatial overlap with in vivo image slices was quantified using the Dice similarity coefficient (DSC).

Results

Imaging of the woodchucks revealed heterogeneous hepatic tumors of varying size, number, and location. Specimen-specific 3D molds provided accurate co-localization of histopathology of whole livers, liver lobes, and pedunculated tumors with in vivo CT and MR imaging, with or without tissue fixation. Visual inspection of histopathology sections and corresponding in vivo image slices revealed spatial registration of analogous pathologic features. The mean DSC for all specimens was 0.83+/-0.05.

Conclusion

Use of specimen-specific 3D molds for en bloc liver dissection provided strong spatial overlap and feature correspondence between in vivo image slices and histopathology sections.

Radiomics for radiation oncologists: are we ready to go?

Radiomics have emerged as an exciting field of research over the past few years, with very wide potential applications in personalised and precision medicine of the future. Radiomics-based approaches are still however limited in daily clinical practice in oncology. This review focus on how radiomics could be incorporated into the radiation therapy pipeline, and globally help the radiation oncologist, from the tumour diagnosis to follow-up after treatment. Radiomics could impact on all steps of the treatment pipeline, once the limitations in terms of robustness and reproducibility are overcome. Major ongoing efforts should be made to collect and share data in the most standardised manner possible.

Normalization of multicenter CT radiomics by a generative adversarial network method

PURPOSE

To reduce the variability of radiomics features caused by computed tomography (CT) imaging protocols through using a generative adversarial network (GAN) method.

MATERIAL AND METHODS

In this study, we defined a set of images acquired with a certain imaging protocol as a domain, and a total of 4 domains (A, B, C, and T [target]) from 3 different scanners were included. In dataset#1, 60 patinets for each domain were collected. Datasets#2 and #3 included 40 slices of spleen for each of the domains. In dataset#4, the slices of 3 colorectal cancer groups (n = 28, 38, and 32) were separately retrieved from 3 different scanners, and each group contained short-term and long-term survivors. 77 features were extracted for evaluation by comparing features distributions. First, we trained the GAN model on dataset#1 to learn how to normalize images from domains A, B, and C to T. Next, by comparing feature distributions between normalized images of the different domains, we identified the appropriate model and assessed it , in dataset #2 and dataset#3, respectively. Finally, to investigate whether our proposed method could facilitate multicenter radiomics analysis, we built the lasso classifier to distinguish short-term from long-term survivors based on a certain group in dataset#4, and validate it in another two groups, which formed a cross-validation between groups in dataset#4.

RESULTS

After normalization, the percentage of aligned features between domains A vs T, B vs T, and C vs T increased from 10.4 %, 18.2%, and 50.1% to 93.5%, 89.6%, and 77.9%, respectively. In the cross-validation results, average improvement of the area under the receiver operating characteristic curve achieved 11% (3%-32%).

CONCLUSION

Our proposed GAN-based normalization method could reduce the variability of radiomics features caused by different CT imaging protocols and facilitate multicenter radiomics analysis.

Development and validation of an individualized nomogram to identify occult peritoneal metastasis in patients with advanced gastric cancer

Background

Occult peritoneal metastasis (PM) in advanced gastric cancer (AGC) patients is highly possible to be missed on computed tomography (CT) images. Patients with occult PMs are subject to late detection or even improper surgical treatment. We therefore aimed to develop a radiomic nomogram to preoperatively identify occult PMs in AGC patients.

Patients and methods

A total of 554 AGC patients from 4 centers were divided into 1 training, 1 internal validation, and 2 external validation cohorts. All patients’ PM status was firstly diagnosed as negative by CT, but later confirmed by laparoscopy (PM-positive n = 122, PM-negative n = 432). Radiomic signatures reflecting phenotypes of the primary tumor (RS1) and peritoneum region (RS2) were built as predictors of PM from 266 quantitative image features. Individualized nomograms of PM status incorporating RS1, RS2, or clinical factors were developed and evaluated regarding prediction ability.

Results

RS1, RS2, and Lauren type were significant predictors of occult PM (all P < 0.05). A nomogram of these three factors demonstrated better diagnostic accuracy than the model with RS1, RS2, or clinical factors alone (all net reclassification improvement P < 0.05). The area under curve yielded was 0.958 [95% confidence interval (CI) 0.923–0.993], 0.941 (95% CI 0.904–0.977), 0.928 (95% CI 0.886–0.971), and 0.920 (95% CI 0.862–0.978) for the training, internal, and two external validation cohorts, respectively. Stratification analysis showed that this nomogram had potential generalization ability.

Conclusion

CT phenotypes of both primary tumor and nearby peritoneum are significantly associated with occult PM status. A nomogram of these CT phenotypes and Lauren type has an excellent prediction ability of occult PM, and may have significant clinical implications on early detection of occult PM for AGC.

CT-based radiomics for preoperative prediction of early recurrent hepatocellular carcinoma: technical reproducibility of acquisition and scanners

PURPOSE

To test the technical reproducibility of acquisition and scanners of CT image-based radiomics model for early recurrent hepatocellular carcinoma (HCC).

METHODS

We included primary HCC patient undergone curative therapies, using early recurrence as endpoint. Four datasets were constructed: 109 images from hospital #1 for training (set 1: 1-mm image slice thickness), 47 images from hospital #1 for internal validation (sets 2 and 3: 1-mm and 10-mm image slice thicknesses, respectively), and 47 images from hospital #2 for external validation (set 4: vastly different from training dataset). A radiomics model was constructed. Radiomics technical reproducibility was measured by overfitting and calibration deviation in external validation dataset. The influence of slice thickness on reproducibility was evaluated in two internal validation datasets.

RESULTS

Compared with set 1, the model in set 2 indicated favorable prediction efficiency (the area under the curve 0.79 vs. 0.80, P = 0.47) and good calibration (unreliability statistic U: P = 0.33). However, in set 4, significant overfitting (0.63 vs. 0.80, P < 0.01) and calibration deviation (U: P < 0.01) were observed. Similar poor performance was also observed in set 3 (0.56 vs. 0.80, P = 0.02; U: P < 0.01).

CONCLUSIONS

CT-based radiomics has poor reproducibility between centers. Image heterogeneity, such as slice thickness, can be a significant influencing factor.

Radiomics model based on preoperative gadoxetic acid-enhanced MRI for predicting liver failure

BACKGROUND

Postoperative liver failure is the most severe complication in cirrhotic patients with hepatocellular carcinoma (HCC) after major hepatectomy. Current available clinical indexes predicting postoperative residual liver function are not sufficiently accurate.

AIM

To determine a radiomics model based on preoperative gadoxetic acid-enhanced magnetic resonance imaging for predicting liver failure in cirrhotic patients with HCC after major hepatectomy.

METHODS

For this retrospective study, a radiomics-based model was developed based on preoperative hepatobiliary phase gadoxetic acid-enhanced magnetic resonance images in 101 patients with HCC between June 2012 and June 2018. Sixty-one radiomic features were extracted from hepatobiliary phase images and selected by the least absolute shrinkage and selection operator method to construct a radiomics signature. A clinical prediction model, and radiomics-based model incorporating significant clinical indexes and radiomics signature were built using multivariable logistic regression analysis. The integrated radiomics-based model was presented as a radiomics nomogram. The performances of clinical prediction model, radiomics signature, and radiomics-based model for predicting post-operative liver failure were determined using receiver operating characteristics curve, calibration curve, and decision curve analyses.

RESULTS

Five radiomics features from hepatobiliary phase images were selected to construct the radiomics signature. The clinical prediction model, radiomics signature, and radiomics-based model incorporating indocyanine green clearance rate at 15 min and radiomics signature showed favorable performance for predicting postoperative liver failure (area under the curve: 0.809-0.894). The radiomics-based model achieved the highest performance for predicting liver failure (area under the curve: 0.894; 95%CI: 0.823-0.964). The integrated discrimination improvement analysis showed a significant improvement in the accuracy of liver failure prediction when radiomics signature was added to the clinical prediction model (integrated discrimination improvement = 0.117, P = 0.002). The calibration curve and an insignificant Hosmer-Lemeshow test statistic (P = 0.841) demonstrated good calibration of the radiomics-based model. The decision curve analysis showed that patients would benefit more from a radiomics-based prediction model than from a clinical prediction model and radiomics signature alone.

CONCLUSION

A radiomics-based model of preoperative gadoxetic acid–enhanced MRI can be used to predict liver failure in cirrhotic patients with HCC after major hepatectomy.

Technology-driven research for radiotherapy innovation

Technology has a pivotal role in the continuous development of radiotherapy. The long road toward modern ‘high‐tech’ radiation oncology has been studded with discoveries and technological innovations that resulted from the interaction of various disciplines. In the last decades, a dramatic technology‐driven revolution has hugely improved the capability of accurately and safely delivering complex‐shaped dose distributions. This has contributed to many clinical improvements, such as the successful management of lung cancer and oligometastatic disease through stereotactic body radiotherapy. Technology‐driven research is an active and lively field with promising potential in several domains, including image guidance, adaptive radiotherapy, integration of artificial intelligence, heavy‐particle therapy, and ‘flash’ ultra‐high dose‐rate radiotherapy. The evolution toward personalized Oncology will deeply influence technology‐driven research, aiming to integrate predictive models and omics analyses into fast and efficient solutions to deliver the best treatment for each single patient. Personalized radiation oncology will need affordable technological solutions for middle‐/low‐income countries, as these are expected to experience the highest increase of cancer incidence and mortality. Moreover, technology solutions for automation of commissioning, quality assurance, safety tests, image segmentation, and plan optimization will be required. Although a large fraction of cancer patients receive radiotherapy, this is certainly not reflected in the worldwide budget for radiotherapy research. Differently from the pharmaceutical companies‐driven research, resources for research in radiotherapy are highly limited to equipment vendors, who can, in turn, initiate a limited number of collaborations with academic research centers. Thus, enhancement of investments in technology‐driven radiotherapy research via public funds, national governments, and the European Union would have a crucial societal impact. It would allow for radiotherapy to further strengthen its role as a highly effective and cost‐efficient cancer treatment modality, and it could facilitate a rapid and equalitarian large‐scale transfer of technology to clinic, with direct impact on patient care.

Noninvasive Fuhrman grading of clear cell renal cell carcinoma using computed tomography radiomic features and machine learning

PURPOSE

To identify optimal classification methods for computed tomography (CT) radiomics-based preoperative prediction of clear cell renal cell carcinoma (ccRCC) grade.

MATERIALS AND METHODS

Seventy-one ccRCC patients (31 low grade and 40 high grade) were included in this study. Tumors were manually segmented on CT images followed by the application of three image preprocessing techniques (Laplacian of Gaussian, wavelet filter, and discretization of the intensity values) on delineated tumor volumes. Overall, 2530 radiomics features (tumor shape and size, intensity statistics, and texture) were extracted from each segmented tumor volume. Univariate analysis was performed to assess the association between each feature and the histological condition. Multivariate analysis involved the use of machine learning (ML) algorithms and the following three feature selection algorithms: the least absolute shrinkage and selection operator, Student's t test, and minimum Redundancy Maximum Relevance. These selected features were then used to construct three classification models (SVM, random forest, and logistic regression) to discriminate high from low-grade ccRCC at nephrectomy. Lastly, multivariate model performance was evaluated on the bootstrapped validation cohort using the area under the receiver operating characteristic curve (AUC) metric.

RESULTS

The univariate analysis demonstrated that among the different image sets, 128 bin-discretized images have statistically significant different texture parameters with a mean AUC of 0.74 ± 3 (q value < 0.05). The three ML-based classifiers showed proficient discrimination between high and low-grade ccRCC. The AUC was 0.78 for logistic regression, 0.62 for random forest, and 0.83 for the SVM model, respectively.

CONCLUSION

CT radiomic features can be considered as a useful and promising noninvasive methodology for preoperative evaluation of ccRCC Fuhrman grades.

Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma: Initial Application of a Radiomic Algorithm Based on Grayscale Ultrasound Images

Objectives: To establish a radiomic algorithm based on grayscale ultrasound images and to make preoperative predictions of microvascular invasion (MVI) in hepatocellular carcinoma (HCC) patients. Methods: In this retrospective study, 322 cases of histopathologically confirmed HCC lesions were included. The classifications based on preoperative grayscale ultrasound images were performed in two stages: (1) classifier #1, MVI-negative and MVI-positive cases; (2) classifier #2, MVI-positive cases were further classified as M1 or M2 cases. The gross-tumoral region (GTR) and peri-tumoral region (PTR) signatures were combined to generate gross- and peri-tumoral region (GPTR) radiomic signatures. The optimal radiomic signatures were further incorporated with vital clinical information. Multivariable logistic regression was used to build radiomic models. Results: Finally, 1,595 radiomic features were extracted from each HCC lesion. At the classifier #1 stage, the radiomic signatures based on features of GTR, PTR, and GPTR showed area under the curve (AUC) values of 0.708 (95% CI, 0.603-0.812), 0.710 (95% CI, 0.609-0.811), and 0.726 (95% CI, 0.625-0.827), respectively. Upon incorporation of vital clinical information, the AUC of the GPTR radiomic algorithm was 0.744 (95% CI, 0.646-0.841). At the classifier #2 stage, the AUC of the GTR radiomic signature was 0.806 (95% CI, 0.667-0.944). Conclusions: Our radiomic algorithm based on grayscale ultrasound images has potential value to facilitate preoperative prediction of MVI in HCC patients. The GTR radiomic signature may be helpful for further discriminating between M1 and M2 levels among MVI-positive patients.

Will traditional biopsy be substituted by radiomics and liquid biopsy for breast cancer diagnosis and characterisation?

The diagnosis of breast cancer currently relies on radiological and clinical evaluation, confirmed by histopathological examination. However, such approach has some limitations as the suboptimal sensitivity, the long turnaround time for recall tests, the invasiveness of the procedure and the risk that some features of target lesions may remain undetected, making re-biopsy a necessity. Recent technological advances in the field of artificial intelligence hold promise in addressing such medical challenges not only in cancer diagnosis, but also in treatment assessment, and monitoring of disease progression. In the perspective of a truly personalised medicine, based on the early diagnosis and individually tailored treatments, two new technologies, namely radiomics and liquid biopsy, are rising as means to obtain information from diagnosis to molecular profiling and response assessment, without the need of a biopsied tissue sample. Radiomics works through the extraction of quantitative peculiar features of cancer from radiological data, while liquid biopsy gets the whole of the malignancy's biology from something as easy as a blood sample. Both techniques hopefully will identify diagnostic and prognostic information of breast cancer potentially reducing the need for invasive (and often difficult to perform) biopsies and favouring an approach that is as personalised as possible for each patient. Nevertheless, such techniques will not substitute tissue biopsy in the near future, and even in further times they will require the aid of other parameters to be correctly interpreted and acted upon.

MRI-based radiogenomics analysis for predicting genetic alterations in oncogenic signalling pathways in invasive breast carcinoma

AIM

To investigate the effect of radiomics in the assessment of alterations in canonical cancer pathways in breast cancer.

MATERIALS AND METHODS

Eighty-eight biopsy-proven breast cancer cases were included in the present study. Radiomics features were extracted from T1-weighted sagittal dynamic contrast-enhanced magnetic resonance imaging (MRI) images. Radiomics signatures were developed to predict genetic alterations in the cell cycle, Myc, PI3K, RTK/RAS, and p53 signalling pathways by using hypothesis testing combined with least absolute shrinkage and selection operator (LASSO) regression analysis. The predictive powers of the models were examined by the area under the curve (AUC) of the receiver operating characteristic curve.

RESULTS

A total of 5,234 radiomics features were obtained from MRI images based on the tumour region of interest. Hypothesis tests screened 250, 229, 156, 785, and 319 radiomics features that were differentially displayed between cell cycle, Myc, PI3K, RTK/RAS, and p53 alterations and no alteration status. According to the LASSO algorithm, 11, 12, 12, 15, and 13 features were identified for the construction of the radiomics signatures to predict cell cycle, Myc, PI3K, RTK/RAS, and p53 alterations, with AUC values of 0.933, 0.926, 0.956, 0.940, and 0.886, respectively. The cell cycle radiomics score correlated closely with the RTK/RAS and p53 radiomics scores. These signatures were also dysregulated in patients with different oestrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 statuses.

CONCLUSION

MRI-based radiogenomics analysis exhibits excellent performance in predicting genetic pathways alterations, thus providing a novel approach for non-invasively obtaining genetic-level molecular characteristics of tumours.

Predicting the Type of Tumor-Related Epilepsy in Patients With Low-Grade Gliomas: A Radiomics Study

Purpose: The majority of patients with low-grade gliomas (LGGs) experience tumor-related epilepsy during the disease course. Our study aimed to build a radiomic prediction model for LGG-related epilepsy type based on magnetic resonance imaging (MRI) data.

Methods: A total of 205 cases with LGG-related epilepsy were enrolled in the retrospective study and divided into training and validation cohorts (1:1) according to their surgery time. Seven hundred thirty-four radiomic features were extracted from T2-weighted imaging, including six location features. Pearson correlation coefficient, univariate area under curve (AUC) analysis, and least absolute shrinkage and selection operator regression were adopted to select the most relevant features for the epilepsy type to build a radiomic signature. Furthermore, a novel radiomic nomogram was developed for clinical application using the radiomic signature and clinical variables from all patients.

Results: Four MRI-based features were selected from the 734 radiomic features, including one location feature. Good discriminative performances were achieved in both training (AUC = 0.859, 95% CI = 0.787–0.932) and validation cohorts (AUC = 0.839, 95% CI = 0.761–0.917) for the type of epilepsy. The accuracies were 80.4 and 80.6%, respectively. The radiomic nomogram also allowed for a high degree of discrimination. All models presented favorable calibration curves and decision curve analyses.

Conclusion: Our results suggested that the MRI-based radiomic analysis may predict the type of LGG-related epilepsy to enable individualized therapy for patients with LGG-related epilepsy.

Radiomics Method for the Differential Diagnosis of Radionecrosis Versus Progression after Fractionated Stereotactic Body Radiotherapy for Brain Oligometastasis

Stereotactic radiotherapy (SRT) is recommended for treatment of brain oligometastasis (BoM) in patients with controlled primary disease. Where contrast enhancement enlargement occurs during follow-up, distinguishing between radionecrosis and progression presents a critical challenge. Without pathological confirmation, decision-making may be inappropriate and delayed. Quantitative imaging features extracted from routinely performed examinations are of interest in potentially addressing this problem. We explored the added value of the radiomics method for the differential diagnosis of these two entities. Twenty patients who received SRT for BoM, from any primary location, were included (8 radionecrosis, 12 progressions, pathologically confirmed). We assessed the clinical relevance of 1,766 radiomics features, extracted using IBEX software, from the first T1-weighted postcontrast magnetic resonance imaging (MRI) after SRT showing a lesion modification. We evaluated seven feature-selection methods and 12 classification methods in terms of respective predictive performance. The classification accuracy was measured using Cohen's kappa after leave-one-out cross-validation. In this work, the best predictive power reached was a Cohen's kappa of 0.68 (overall accuracy of 85%), expressing a strong agreement between the algorithm prediction and the histological gold standard. Prediction accuracy was 75% for radionecrosis, and 91% for progression. The area under a curve reached 0.83 using a bagging algorithm trained with the chi-square score features set. These findings indicated that the radiomics method is able to discriminate radionecrosis from progression in an accurate, early and noninvasive way. This promising study is a proof of concept, preceding a larger prospective study for defining a robust model to support decision-making in BoM. In summary, distinguishing between radionecrosis and progression is challenging without pathology. We built a classification model based on imaging data and machine learning. Using this model, we were able predict progression and radionecrosis in, respectively, 91% and 75% of cases.

Privacy-preserving distributed learning of radiomics to predict overall survival and HPV status in head and neck cancer

A major challenge in radiomics is assembling data from multiple centers. Sharing data between hospitals is restricted by legal and ethical regulations. Distributed learning is a technique, enabling training models on multicenter data without data leaving the hospitals ("privacy-preserving" distributed learning). This study tested feasibility of distributed learning of radiomics data for prediction of two year overall survival and HPV status in head and neck cancer (HNC) patients. Pretreatment CT images were collected from 1174 HNC patients in 6 different cohorts. 981 radiomic features were extracted using Z-Rad software implementation. Hierarchical clustering was performed to preselect features. Classification was done using logistic regression. In the validation dataset, the receiver operating characteristics (ROC) were compared between the models trained in the centralized and distributed manner. No difference in ROC was observed with respect to feature selection. The logistic regression coefficients were identical between the methods (absolute difference <10-7). In comparison of the full workflow (feature selection and classification), no significant difference in ROC was found between centralized and distributed models for both studied endpoints (DeLong p > 0.05). In conclusion, both feature selection and classification are feasible in a distributed manner using radiomics data, which opens new possibility for training more reliable radiomics models.

Exploring technical issues in personalized medicine: NSCLC survival prediction by quantitative image analysis-usefulness of density correction of volumetric CT data

The aim of this study was to apply density correction method to the quantitative image analysis of non-small cell lung cancer (NSCLC) computed tomography (CT) images, determining its influence on overall survival (OS) prediction of surgically treated patients. Clinicopathological (CP) data and preoperative CT scans, pre- and post-contrast medium (CM) administration, of 57 surgically treated NSCLC patients, were retrospectively collected. After CT volumetric density measurement of primary gross tumour volume (GTV), aorta and tracheal air, density correction was conducted on GTV (reference values: aortic blood and tracheal air). For each resulting data set (combining CM administration and normalization), first-order statistical and textural features were extracted. CP and imaging data were correlated with patients 1-, 3- and 5-year OS, alone and combined (uni-/multivariate logistic regression and Akaike information criterion). Predictive performance was evaluated using the ROC curves and AUC values and compared among non-normalized/normalized data sets (DeLong test). The best predictive values were obtained when combining CP and imaging parameters (AUC values: 1 year 0.72; 3 years 0.82; 5 years 0.78). After normalization resulted an improvement in predicting 1-year OS for some of the grey level size zonebased features (large zone low grey level emphasis) and for the combined CP-imaging model, a worse performance for grey level co-occurrence matrix (cluster prominence and shade) and first-order statistical (range) parameters for 1- and 5-year OS, respectively. The negative performance of cluster prominence in predicting 1-year OS was the only statistically significant result (p value 0.05). Density corrections of volumetric CT data showed an opposite influence on the performance of imaging quantitative features in predicting OS of surgically treated NSCLC patients, even if no statistically significant for almost all predictors.

Application of CT-based radiomics in predicting portal pressure and patient outcome in portal hypertension

PURPOSE

Portal venous pressure (PVP) measurement is of clinical significance, especially in patients with portal hypertension. However, the invasive nature and associated complications limits its application. The aim of the study is to propose a noninvasive predictive model of PVP values based on CT-extracted radiomic features.

METHODS

Radiomics PVP (rPVP) models based on liver, spleen and combined features were established on an experimental cohort of 169 subjects. Radiomics features were extracted from each ROI and reduced via the LASSO regression to achieve an optimal predictive formula. A validation cohort of 62 patients treated for gastroesophageal varices (GOV) was used to confirm the utility of rPVP in predicting variceal recurrence. The association between rPVP and response to treatment was observed.

RESULTS

Three separate predictive formula for PVP were derived from radiomics features. rPVP was significantly correlated to patient response to endoscopic treatment for GOV. Among which, the model containing both liver and spleen features has the highest predictability of variceal recurrence, with an optimal cut-off value at 29.102 mmHg (AUC 0.866). A Kaplan Meier analysis further confirmed the difference between patients with varying rPVP values.

CONCLUSION

PVP values can be accurately predicted by a non-invasive, CT derived radiomics model. rPVP serves as a non-invasive and precise reference for predicting treatment outcome for GOV secondary to portal hypertension.

An ensemble learning approach for brain cancer detection exploiting radiomic features

BACKGROUND AND OBJECTIVE

The brain cancer is one of the most aggressive tumour: the 70% of the patients diagnosed with this malignant cancer will not survive. Early detection of brain tumours can be fundamental to increase survival rates. The brain cancers are classified into four different grades (i.e., I, II, III and IV) according to how normal or abnormal the brain cells look. The following work aims to recognize the different brain cancer grades by analysing brain magnetic resonance images.

METHODS

A method to identify the components of an ensemble learner is proposed. The ensemble learner is focused on the discrimination between different brain cancer grades using non invasive radiomic features. The considered radiomic features are belonging to five different groups: First Order, Shape, Gray Level Co-occurrence Matrix, Gray Level Run Length Matrix and Gray Level Size Zone Matrix. We evaluate the features effectiveness through hypothesis testing and through decision boundaries, performance analysis and calibration plots thus we select the best candidate classifiers for the ensemble learner.

RESULTS

We evaluate the proposed method with 111,205 brain magnetic resonances belonging to two freely available data-sets for research purposes. The results are encouraging: we obtain an accuracy of 99% for the benign grade I and the II, III and IV malignant brain cancer detection.

CONCLUSION

The experimental results confirm that the ensemble learner designed with the proposed method outperforms the current state-of-the-art approaches in brain cancer grade detection starting from magnetic resonance images.

Circulating biomarkers of systemic sclerosis - interstitial lung disease

Interstitial lung disease is a frequent organ manifestation in systemic sclerosis and is associated with high mortality. It is crucial to diagnose interstitial lung disease in systemic sclerosis and to assess severity and identify patients prone to progression at an early stage to ultimately decrease organ damage and improve outcome. Circulating anti-topoisomerase-I autoantibodies have long been associated with the presence and development of systemic sclerosis - interstitial lung disease, evidence on their potential to further predict the clinical course of systemic sclerosis is however conflicting. C-reactive protein is a marker of infection and systemic inflammation with widespread clinical application and is elevated in systemic sclerosis with a tendency towards higher abundancy in patients with early disease. The role of other circulating biomarkers is promising but hampered by the lack of standardized criteria and guidelines for sample/data collection, analyses, reporting and validation and has not reached prime time for clinical application. However, epithelial markers including Krebs von den Lungen-6 and surfactant protein D and several cytokines and chemokines including CCL2 and CCL18 for severity assessment of systemic sclerosis - interstitial lung disease patients at the time of interstitial lung disease diagnosis and to predict interstitial lung disease progression have been reported and seem to be promising candidate biomarkers in the future.

Development and validation of a radiomics model based on T2WI images for preoperative prediction of microsatellite instability status in rectal cancer: Study Protocol Clinical Trial (SPIRIT Compliant)

INTRODUCTION

Globally, colorectal cancer (CRC) is the third most commonly diagnosed cancer in males and the second in females. Rectal cancer (RC) accounts for about 28% of all newly diagnosed CRC cases. The treatment of choice for locally advanced RC is a combination of surgical resection and chemotherapy and/or radiotherapy. These patients can potentially be cured, but the clinical outcome depends on the tumor biology. Microsatellite instability (MSI) is an important biomarker in CRC, with crucial diagnostic, prognostic, and predictive implications. It is important to develop a noninvasive, repeatable, and reproducible method to reflect the microsatellite status. Magnetic resonance imaging (MRI) has been recommended as the preferred imaging examination for RC in clinical practice by both the National Comprehensive Cancer Network and the European Society for Medical Oncology guidelines. T2WI is the core sequence of MRI scanning protocol for RC. Radiomics, the high-throughput mining of quantitative image features from standard-of-care medical imaging that enables data to be extracted and applied within clinical-decision support systems to improve diagnostic, prognostic, and predictive accuracy, is gaining importance in cancer research.We proposed a hypothesis: A simple radiomics model based on only T2WI images can accurately evaluate the MSI status of RC preoperatively.

OBJECTIVE

To develop a radiomics model based on T2WI images for accurate preoperative diagnosis the MSI status of RC.

METHOD

All patients with RC were retrospectively enrolled. The dataset was randomly split into training cohort (70% of all patients) and testing cohort (30% of all patients). The radiomics features will be extracted from T2WI-MR images of the entire primary tumor region. Least absolute shrinkage and selection operator was used to select the most predictive radiomics features. Logistic regression models were constructed in the training/validation cohort to discriminate the MSI status using clinical factors, radiomics features, or their integration. The diagnostic performance of these 3 models was evaluated in the testing cohort based on their area under the curve, sensitivity, specificity, and accuracy.

DISCUSSION

This study will help us know whether radiomics model based on T2WI images to preoperative identify MSI status of RC.

Complementary role of computed tomography texture analysis for differentiation of pancreatic ductal adenocarcinoma from pancreatic neuroendocrine tumors in the portal-venous enhancement phase

Purpose

To assess the role of CT-texture analysis (CTTA) for differentiation of pancreatic ductal adenocarcinoma (PDAC) from pancreatic neuroendocrine neoplasm (PNEN) in the portal-venous phase as compared with visual assessment and tumor-to-pancreas attenuation ratios.

Methods

53 patients (66.1 ± 8.6y) with PDAC and 42 patients (65.5 ± 12.2y) with PNEN who underwent contrast-enhanced CT for primary staging were evaluated. Volumes of interests (VOIs) were set in the tumor tissue at the portal-venous phase excluding adjacent structures. Based on pyradiomics library, 92 textural features were extracted including 1st, 2nd, and higher order features, and then compared between PNEN and PDAC. The visual assessment classified tumors into hypo-, iso-, or hyperdense to pancreas parenchyma or into homogeneous/heterogeneous. Additionally, attenuation ratios between the tumors and the non-involved pancreas were calculated.

Results

8/92 (8.6%) highly significant (p < 0.005) discriminatory textural features between PDAC and PNEN were identified including the 1st order features “median,” “total energy,” “energy,” “10th percentile,” “90th percentile,” “minimum,” “maximum,” and the 2nd order feature “Gray-Level co-occurrence Matrix (GLCM) Informational Measure of Correlation (Imc2).” In PNEN, the higher order feature “GLSZM Small Area High Gray-Level Emphasis” proved significantly higher in G1 compared to G2/3 tumors (p < 0.05). The tumor/parenchyma ratios as well as the visual assessment into hypo-/iso-/hyperdense or homogeneous/heterogeneous did not significantly differ between PDAC and PNEN.

Conclusions

Our data indicate that CTTA is a feasible tool for differentiation of PNEN from PDAC and also of G1 from G2/3 PNEN in the portal-venous phase. Visual assessment and tumor-to-parenchyma ratios were not useful for discrimination.

Electronic supplementary material

The online version of this article (10.1007/s00261-020-02406-9) contains supplementary material, which is available to authorized users.

CT-based radiomics and machine learning to predict spread through air space in lung adenocarcinoma

PURPOSE

Spread through air space (STAS) is a novel invasive pattern of lung adenocarcinoma and is also a risk factor for recurrence and worse prognosis of lung adenocarcinoma. The aims of this study are to develop and validate a computed tomography (CT)‑based radiomics model for preoperative prediction of STAS in lung adenocarcinoma.

METHODS AND MATERIALS

This retrospective study was approved by an institutional review board and included 462 (mean age, 58.06 years) patients with pathologically confirmed lung adenocarcinoma. STAS was identified in 90 patients (19.5%). Two experienced radiologists segmented and extracted radiomics features on preoperative thin-slice CT images with radiomics extension independently. Intraclass correlation coefficients (ICC) and Pearson's correlation were used to rule out those low reliable (ICC < 0.75) and redundant (r > 0.9) features. Univariate logistic regression was applied to select radiomics features which were associated with STAS. A radiomics-based machine learning predictive model using a random forest (RF) was developed and calibrated with fivefold cross-validation. The diagnostic performance of the model was measured by the area under the curve (AUC) of receiver operating characteristic (ROC).

RESULTS

With univariate analysis, 12 radiomics features and age were found to be associated with STAS significantly. The RF model achieved an AUC of 0.754 (a sensitivity of 0.880 and a specificity of 0.588) for predicting STAS.

CONCLUSION

CT-based radiomics model can preoperatively predict STAS in lung adenocarcinoma with good diagnosis performance.

KEY POINTS

• CT-based radiomics and machine learning model can predict spread through air space (STAS) in lung adenocarcinoma with high accuracy. • The random forest (RF) model achieved an AUC of 0.754 (a sensitivity of 0.880 and a specificity of 0.588) for predicting STAS.

A Contrast-Enhanced Computed Tomography Based Radiomics Approach for Preoperative Differentiation of Pancreatic Cystic Neoplasm Subtypes: A Feasibility Study

Background: Serous cystadenoma (SCA), mucinous cystadenoma (MCN), and intraductal papillary mucinous neoplasm (IPMN) are three subtypes of pancreatic cystic neoplasm (PCN). Due to the potential of malignant-transforming, patients with MCN and IPMN require radical surgery while patients with SCA need periodic surveillance. However, accurate pre-surgery diagnosis between SCA, MCN, and IPMN remains challenging in the clinic. Methods: This study enrolled 164 patients including 76 with SCA, 40 with MCN and 48 with IPMN. Patients were randomly split into a training cohort (n = 115) and validation cohort (n = 41). We performed statistical analysis and Boruta method to screen significantly distinct clinical factors and radiomics features extracted on pre-surgery contrast-enhanced computed tomography (CECT) images among three subtypes. Three reliable machine-learning algorithms, support vector machine (SVM), random forest (RF) and artificial neural network (ANN), were utilized to construct classifiers based on important radiomics features and clinical parameters. Precision, recall, and F1-score were calculated to assess the performance of the constructed classifiers. Results: Nine of 547 radiomics features and eight clinical factors showed a significant difference among SCA, MCN, and IPMN. Five radiomics features (Histogram_Entropy, Histogram_Skeweness, LLL_GLSZM_GLV, Histogram_Uniformity, HHL_Histogram_Kurtosis), and four clinical factors, including serum carbohydrate antigen 19-9, sex, age, and serum carcinoembryonic antigen, were identified important by Boruta method. The SVM classifier achieved an overall accuracy of 73.04% in training cohort and 71.43% in validation cohort, respectively. The RF classifier achieved overall accuracy of 84.35 and 79.59%, respectively. The constructed ANN model showed an overall accuracy of 77.39% in the training dataset and 71.43% in the validation dataset. All the three classifiers showed high F1 score for differentiation among the three subtypes. Conclusion: Our study proved the feasibility and translational value of CECT-based radiomics classifiers for differentiation among SCA, MCN, and IPMN.