Performances and variability of CT radiomics for the prediction of microvascular invasion and survival in patients with HCC: a matter of chance or standardisation?

Combining radiomics and imaging biomarkers with clinical variables for the prediction of HCC recurrence after liver transplantation

To develop and validate an integrated model that combines CT-based radiomics and imaging biomarkers with clinical variables to predict recurrence and recurrence-free survival in patients with HCC following liver transplantation (LT), this 2-center retrospective study includes 123 patients with HCC who underwent LT between 2007 and 2021. Radiomic features (RFs) were extracted from baseline CT liver tumor volume. Feature selection was performed using the Least Absolute Shrinkage and Selection Operator (LASSO) regression method with 10-fold cross-validation in the training cohort (n=48) to build a predictive radiomics signature for HCC recurrence. Combined diagnostic models were built based on the radiomics signature supplemented with imaging features beyond the Milan criteria, the AFP (alpha-fetoprotein) model, and Metroticket 2.0 before LT using multivariate logistic regression. Receiver operating characteristic analyses were performed in both internal (n=22) and external (n=53) validation cohorts, and patients were stratified into either high-risk or low-risk groups for HCC recurrence. Kaplan-Meier analysis was performed to analyze recurrence-free survival. LASSO and multivariate regression analysis revealed 4 independent predictors associated with an increased risk of HCC recurrence: radiomics signature of 5 RF, peritumoral enhancement, satellite nodules, and no bridging therapies. For the prediction of tumor recurrence, the highest AUC of the final integrated models combining clinical variables, non-radiomics imaging features, and radiomics was 0.990 and 0.900 for the internal and external validation sets, respectively, outperforming the Milan and clinical stand-alone models. In all integrated models, the high-risk groups had a shorter recurrence-free survival than the corresponding low-risk group. CT-based radiomics and imaging parameters beyond the Milan criteria representing aggressive behavior, along with the history of bridging therapies, show potential for predicting HCC recurrence after LT.

Integrating CT Radiomics and Clinical Features to Optimize TACE Technique Decision-Making in Hepatocellular Carcinoma

BACKGROUND/OBJECTIVES

To develop a decision framework integrating computed tomography (CT) radiomics and clinical factors to guide the selection of transarterial chemoembolization (TACE) technique for optimizing treatment response in non-resectable hepatocellular carcinoma (HCC).

METHODS

A retrospective analysis was performed on 151 patients [33 conventional TACE (cTACE), 69 drug-eluting bead TACE (DEB-TACE), 49 degradable starch microsphere TACE (DSM-TACE)] who underwent TACE for HCC at a single tertiary center. Pre-TACE contrast-enhanced CT images were used to extract radiomic features of the TACE-treated liver tumor volume. Patient clinical and laboratory data were combined with radiomics-derived predictors in an elastic net regularized logistic regression model to identify independent factors associated with early response at 4-6 weeks post-TACE. Predicted response probabilities under each TACE technique were compared with the actual techniques performed.

RESULTS

Elastic net modeling identified three independent predictors of response: radiomic feature "Contrast" (OR = 5.80), BCLC stage B (OR = 0.92), and viral hepatitis etiology (OR = 0.74). Interaction models indicated that the relative benefit of each TACE technique depended on the identified patient-specific predictors. Model-based recommendations differed from the actual treatment selected in 66.2% of cases, suggesting potential for improved patient-technique matching.

CONCLUSIONS

Integrating CT radiomics with clinical variables may help identify the optimal TACE technique for individual HCC patients. This approach holds promise for a more personalized therapy selection and improved response rates beyond standard clinical decision-making.

ESR Essentials: radiomics-practice recommendations by the European Society of Medical Imaging Informatics

Radiomics is a method to extract detailed information from diagnostic images that cannot be perceived by the naked eye. Although radiomics research carries great potential to improve clinical decision-making, its inherent methodological complexities make it difficult to comprehend every step of the analysis, often causing reproducibility and generalizability issues that hinder clinical adoption. Critical steps in the radiomics analysis and model development pipeline-such as image, application of image filters, and selection of feature extraction parameters-can greatly affect the values of radiomic features. Moreover, common errors in data partitioning, model comparison, fine-tuning, assessment, and calibration can reduce reproducibility and impede clinical translation. Clinical adoption of radiomics also requires a deep understanding of model explainability and the development of intuitive interpretations of radiomic features. To address these challenges, it is essential for radiomics model developers and clinicians to be well-versed in current best practices. Proper knowledge and application of these practices is crucial for accurate radiomics feature extraction, robust model development, and thorough assessment, ultimately increasing reproducibility, generalizability, and the likelihood of successful clinical translation. In this article, we have provided researchers with our recommendations along with practical examples to facilitate good research practices in radiomics. KEY POINTS: Radiomics' inherent methodological complexity should be understood to ensure rigorous radiomic model development to improve clinical decision-making. Adherence to radiomics-specific checklists and quality assessment tools ensures methodological rigor. Use of standardized radiomics tools and best practices enhances clinical translation of radiomics models.

18F-FDG PET/CT-based intratumoral and peritumoral radiomics combining ensemble learning for prognosis prediction in hepatocellular carcinoma: a multi-center study

BACKGROUND

Radiomic models combining intratumoral with peritumoral features are potentially beneficial to enhance the predictive performance. This study aimed to identify the optimal 18F-FDG PET/CT-derived radiomic models for prediction of prognosis in hepatocellular carcinoma (HCC).

METHODS

A total of 135 HCC patients from two institutions were retrospectively included. Four peritumoral regions were defined by dilating tumor region with thicknesses of 2 mm, 4 mm, 6 mm, and 8 mm, respectively. Based on segmentation of intratumoral, peritumoral and integrated volume of interest (VOI), corresponding radiomic features were extracted respectively. After feature selection, a total of 15 intratumoral radiomic models were constructed based on five ensemble learning algorithms and radiomic features from three image modalities. Then, the optimal combination of ensemble learning algorithms and image modality in the intratumoral models was selected to develop subsequent peritumoral radiomic models and integrated radiomic models. Finally, a nomogram was developed incorporating the optimal radiomic model with clinical independent predictors to achieve an intuitive representation of the prediction model.

RESULTS

Among the intratumoral radiomic models, the one which combined PET/CT-based radiomic features with SVM classifier outperformed other models. With the addition of peritumoral information, the integrated model based on an integration of intratumoral and 2 mm-peritumoral VOI, was finally approved as the optimal radiomic model with a mean AUC of 0.831 in the internal validation, and a highest AUC of 0.839 (95%CI:0.718-0.960) in the external test. Furthermore, a nomogram incorporating the optimal radiomic model with HBV infection and TNM status, was able to predict the prognosis for HCC with an AUC of 0.889 (95%CI: 0.799-0.979).

CONCLUSIONS

The integrated intratumoral and peritumoral radiomic model, especially for a 2 mm peritumoral region, was verified as the optimal radiomic model to predict the overall survival of HCC. Furthermore, combination of integrated radiomic model with significant clinical parameter contributed to further enhance the prediction efficacy.

TRIAL REGISTRATION

This study was a retrospective study, so it was free from registration.

A Nomogram Based on MRI Visual Decision Tree to Evaluate Vascular Endothelial Growth Factor in Hepatocellular Carcinoma

BACKGROUNDS

Anti-vascular endothelial growth factor (VEGF) therapy has been developed and recognized as an effective treatment for hepatocellular carcinoma (HCC). However, there remains a lack of noninvasive methods in precisely evaluating VEGF expression in HCC.

PURPOSE

To establish a visual noninvasive model based on clinical indicators and MRI features to evaluate VEGF expression in HCC.

STUDY TYPE

Retrospective.

POPULATION

One hundred forty HCC patients were randomly divided into a training (N = 98) and a test cohort (N = 42).

FIELD STRENGTH/SEQUENCE

3.0 T, T2WI, T1WI including pre-contrast, dynamic, and hepatobiliary phases.

ASSESSMENT

The fusion model constructed by history of smoking, albumin-to-globulin ratio (AGR) and the Radio-Tree model was visualized by a nomogram.

STATISTICAL TESTS

Performances of models were assessed by receiver operating characteristic (ROC) curves. Student's t-test, Mann-Whitney U-test, chi-square test, Fisher's exact test, univariable and multivariable logistic regression analysis, DeLong's test, integrated discrimination improvement (IDI), Hosmer-Lemeshow test, and decision curve analysis were performed. P < 0.05 was considered statistically significant.

RESULTS

History of smoking and AGR ≤1.5 were clinical independent risk factors of the VEGF expression. In training cohorts, values of area under the curve (AUCs) of Radio-Tree model, Clinical-Radiological (C-R) model, fusion model which combined history of smoking and AGR with Radio-Tree model were 0.821, 0.748, and 0.871. In test cohort, the fusion model showed highest AUC (0.844) than Radio-Tree and C-R models (0.819, 0.616, respectively). DeLong's test indicated that the fusion model significantly differed in performance from the C-R model in training cohort (P = 0.015) and test cohort (P = 0.007).

DATA CONCLUSION

The fusion model combining history of smoking, AGR and Radio-Tree model established with ML algorithm showed the highest AUC value than others.

EVIDENCE LEVEL

4 TECHNICAL EFFICACY: Stage 2.

Delta-radiomics Analysis Based on Multi-phase Contrast-enhanced MRI to Predict Early Recurrence in Hepatocellular Carcinoma After Percutaneous Thermal Ablation

RATIONALE AND OBJECTIVES

It is critical to predict early recurrence (ER) after percutaneous thermal ablation (PTA) for hepatocellular carcinoma (HCC). We aimed to develop and validate a delta-radiomics nomogram based on multi-phase contrast-enhanced magnetic resonance imaging (MRI) to preoperatively predict ER of HCC after PTA.

MATERIALS AND METHODS

We retrospectively enrolled 164 patients with HCC and divided them into training, temporal validation, and other-scanner validation cohorts (n = 110, 29, and 25, respectively). The volumes of interest of the intratumoral and/or peritumoral regions were delineated on preoperative multi-phase MR images. Original radiomics features were extracted from each phase, and delta-radiomics features were calculated. Logistic regression was used to train the corresponding radiomics models. The clinical and radiological characteristics were evaluated and combined to establish a clinical-radiological model. A fusion model comprising the best radiomics scores and clinical-radiological risk factors was constructed and presented as a nomogram. The performance of each model was evaluated and recurrence-free survival (RFS) was assessed.

RESULTS

Child-Pugh grade B, high-risk tumor location, and an incomplete/absent tumor capsule were independent predictors of ER. The optimal radiomics model comprised 12 delta-radiomics features with areas under the curve (AUCs) of 0.834, 0.795, and 0.769 in the training, temporal validation, and other-scanner validation cohorts, respectively. The nomogram showed the best predictive performance with AUCs as 0.893, 0.854, and 0.827 in the three datasets. There was a statistically significant difference in RFS between the risk groups calculated using the delta-radiomics model and nomogram.

CONCLUSIONS

The nomogram combined with the delta-radiomic score and clinical-radiological risk factors could non-invasively predict ER of HCC after PTA.

18F-FDG PET/CT-based habitat radiomics combining stacking ensemble learning for predicting prognosis in hepatocellular carcinoma: a multi-center study

BACKGROUND

This study aims to develop habitat radiomic models to predict overall survival (OS) for hepatocellular carcinoma (HCC), based on the characterization of the intratumoral heterogeneity reflected in 18F-FDG PET/CT images.

METHODS

A total of 137 HCC patients from two institutions were retrospectively included. First, intratumoral habitats were achieved by a two-step unsupervised clustering process based on k-means clustering. Second, a total of 4032 radiomic features were extracted based on each habitat, including 2016 PET-based and 2016 CT-based radiomic features. Then, after feature selection, the stacking ensemble learning approach which combined six machine learning classifiers as the first-level learners with Cox proportional hazards regression as the second-level learner, was employed to build multiple radiomic models. Finally, the optimal model was selected based on the calculation of the C-index, and a combined model integrating with a clinical model was also constructed to identify the potentially complementary effect.

RESULTS

Three spatially distinct habitats were identified in the two cohorts. Among a total of 30 stacking ensemble learning models established based on different combinations of 5 types of segmented volumes of interest (VOIs) with 6 types of classifiers, the MLP-Cox-habitat-2 model was selected as the optimal radiomic model with a C-index of 0.702 in the external validation cohort. Furthermore, the combined model integrating the optimal radiomic model with the clinical model achieved an improved C-index of 0.747. Consistently, the combined model outperformed the other models for OS prediction, with a time-dependent AUC of 0.835, 0.828, and 0.800 in the 1-year, 2-year, and 3-year OS, respectively.

CONCLUSION

18F-FDG PET/CT-based habitat radiomics outperformed traditional radiomics in OS prediction for HCC, with a further improved predictive power by integrating with the clinical model. The optimal combined habitat model was potentially promising in guiding individualized treatment for HCC.

TRIAL REGISTRATION

This study was a retrospective study, so it was free from registration.

radMLBench: A dataset collection for benchmarking in radiomics

BACKGROUND

New machine learning methods and techniques are frequently introduced in radiomics, but they are often tested on a single dataset, which makes it challenging to assess their true benefit. Currently, there is a lack of a larger, publicly accessible dataset collection on which such assessments could be performed. In this study, a collection of radiomics datasets with binary outcomes in tabular form was curated to allow benchmarking of machine learning methods and techniques.

METHODS

A variety of journals and online sources were searched to identify tabular radiomics data with binary outcomes, which were then compiled into a homogeneous data collection that is easily accessible via Python. To illustrate the utility of the dataset collection, it was applied to investigate whether feature decorrelation prior to feature selection could improve predictive performance in a radiomics pipeline.

RESULTS

A total of 50 radiomic datasets were collected, with sample sizes ranging from 51 to 969 and 101 to 11165 features. Using this data, it was observed that decorrelating features did not yield any significant improvement on average.

CONCLUSIONS

A large collection of datasets, easily accessible via Python, suitable for benchmarking and evaluating new machine learning techniques and methods was curated. Its utility was exemplified by demonstrating that feature decorrelation prior to feature selection does not, on average, lead to significant performance gains and could be omitted, thereby increasing the robustness and reliability of the radiomics pipeline.

Bi-regional dynamic contrast-enhanced MRI for prediction of microvascular invasion in solitary BCLC stage A hepatocellular carcinoma

OBJECTIVES

To construct a combined model based on bi-regional quantitative dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), as well as clinical-radiological (CR) features for predicting microvascular invasion (MVI) in solitary Barcelona Clinic Liver Cancer (BCLC) stage A hepatocellular carcinoma (HCC), and to assess its ability for stratifying the risk of recurrence after hepatectomy.

METHODS

Patients with solitary BCLC stage A HCC were prospective collected and randomly divided into training and validation sets. DCE perfusion parameters were obtained both in intra-tumoral region (ITR) and peritumoral region (PTR). Combined DCE perfusion parameters (CDCE) were constructed to predict MVI. The combined model incorporating CDCE and CR features was developed and evaluated. Kaplan-Meier method was used to investigate the prognostic significance of the model and the survival benefits of different hepatectomy approaches.

RESULTS

A total of 133 patients were included. Total blood flow in ITR and arterial fraction in PTR exhibited the best predictive performance for MVI with areas under the curve (AUCs) of 0.790 and 0.792, respectively. CDCE achieved AUCs of 0.868 (training set) and 0.857 (validation set). A combined model integrated with the α-fetoprotein, corona enhancement, two-trait predictor of venous invasion, and CDCE could improve the discrimination ability to AUCs of 0.966 (training set) and 0.937 (validation set). The combined model could stratify the prognosis of HCC patients. Anatomical resection was associated with a better prognosis in the high-risk group (p < 0.05).

CONCLUSION

The combined model integrating DCE perfusion parameters and CR features could be used for MVI prediction in HCC patients and assist clinical decision-making.

CRITICAL RELEVANCE STATEMENT

The combined model incorporating bi-regional DCE-MRI perfusion parameters and CR features predicted MVI preoperatively, which could stratify the risk of recurrence and aid in optimizing treatment strategies.

KEY POINTS

Microvascular invasion (MVI) is a significant predictor of prognosis for hepatocellular carcinoma (HCC). Quantitative DCE-MRI could predict MVI in solitary BCLC stage A HCC; the combined model improved performance. The combined model could help stratify the risk of recurrence and aid treatment planning.

Application of artificial intelligence radiomics in the diagnosis, treatment, and prognosis of hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer, with an increasing incidence and poor prognosis. In the past decade, artificial intelligence (AI) technology has undergone rapid development in the field of clinical medicine, bringing the advantages of efficient data processing and accurate model construction. Promisingly, AI-based radiomics has played an increasingly important role in the clinical decision-making of HCC patients, providing new technical guarantees for prediction, diagnosis, and prognostication. In this review, we evaluated the current landscape of AI radiomics in the management of HCC, including its diagnosis, individual treatment, and survival prognosis. Furthermore, we discussed remaining challenges and future perspectives regarding the application of AI radiomics in HCC.

Radiomics and liver: Where we are and where we are headed?

Hepatic diffuse conditions and focal liver lesions represent two of the most common scenarios to face in everyday radiological clinical practice. Thanks to the advances in technology, radiology has gained a central role in the management of patients with liver disease, especially due to its high sensitivity and specificity. Since the introduction of computed tomography (CT) and magnetic resonance imaging (MRI), radiology has been considered the non-invasive reference modality to assess and characterize liver pathologies. In recent years, clinical practice has moved forward to a quantitative approach to better evaluate and manage each patient with a more fitted approach. In this setting, radiomics has gained an important role in helping radiologists and clinicians characterize hepatic pathological entities, in managing patients, and in determining prognosis. Radiomics can extract a large amount of data from radiological images, which can be associated with different liver scenarios. Thanks to its wide applications in ultrasonography (US), CT, and MRI, different studies were focused on specific aspects related to liver diseases. Even if broadly applied, radiomics has some advantages and different pitfalls. This review aims to summarize the most important and robust studies published in the field of liver radiomics, underlying their main limitations and issues, and what they can add to the current and future clinical practice and literature.

Radiomics in CT and MR imaging of the liver and pancreas: tools with potential for clinical application

Radiomics allows the extraction of quantitative imaging features from clinical magnetic resonance imaging (MRI) and computerized tomography (CT) studies. The advantages of radiomics have primarily been exploited in oncological applications, including better characterization and staging of oncological lesions and prediction of patient outcomes and treatment response. The potential introduction of radiomics in the clinical setting requires the establishment of a standardized radiomics pipeline and a quality assurance program. Radiomics and texture analysis of the liver have improved the differentiation of hypervascular lesions such as adenomas, focal nodular hyperplasia, and hepatocellular carcinoma (HCC) during the arterial phase, and in the pretreatment determination of HCC prognostic factors (e.g., tumor grade, microvascular invasion, Ki-67 proliferation index). Radiomics of pancreatic CT and MR images has enhanced pancreatic ductal adenocarcinoma detection and its differentiation from pancreatic neuroendocrine tumors, mass-forming chronic pancreatitis, or autoimmune pancreatitis. Radiomics can further help to better characterize incidental pancreatic cystic lesions, accurately discriminating benign from malignant intrapancreatic mucinous neoplasms. Nonetheless, despite their encouraging results and exciting potential, these tools have yet to be implemented in the clinical setting. This non-systematic review will describe the essential steps in the implementation of the radiomics and feature extraction workflow from liver and pancreas CT and MRI studies for their potential clinical application. A succinct overview of reported radiomics applications in the liver and pancreas and the challenges and limitations of their implementation in the clinical setting is also discussed, concluding with a brief exploration of the future perspectives of radiomics in the gastroenterology field.

Hepatocellular Carcinoma: Optimal Radiological Evaluation before Liver Transplantation

Liver transplantation (LT) is the recommended curative-intent treatment for patients with early or intermediate-stage hepatocellular carcinoma (HCC) who are ineligible for resection. Imaging plays a central role in staging and for selecting the best LT candidates. This review will discuss recent developments in pre-LT imaging assessment, in particular LT eligibility criteria on imaging, the technical requirements and the diagnostic performance of imaging for the pre-LT diagnosis of HCC including the recent Liver Imaging Reporting and Data System (LI-RADS) criteria, the evaluation of the response to locoregional therapy, as well as the non-invasive prediction of HCC aggressiveness and its impact on the outcome of LT. We will also briefly discuss the role of nuclear medicine in the pre-LT evaluation and the emerging role of artificial intelligence models in patients with HCC.