Though several clinicopathological features are identified as prognostic indicators, potentially prognostic radiomic models are expected to preoperatively and noninvasively predict survival for HCC. Traditional radiomic models are lacking in a consideration for intratumoral regional heterogeneity. The study aimed to establish and validate the predictive power of multiple habitat radiomic models in predicting prognosis of hepatocellular carcinoma (HCC). A total of 232 HCC patients were retrospectively included, including a training/validation cohort and two external testing cohorts from 4 centers. For habitat radiomics, intratumoral habitat partitioning based on CT images was first performed by using Otsu thresholding method. Second, a total of 350 habitat radiomic models were constructed to select the optimal model. Then, both ROC curve analyses and Kaplan-Meier survival curve analyses were applied to assess the predictive performances. Ultimately, an immune status profiling was conducted based on bioinformatic analyses and multiplex immunohistochemistry (mIHC) assays to reveal the potential mechanisms. A total of 4 habitats were segmented, and the corresponding habitat radiomic models were constructed based on each habitat and an integration of all the four habitats. Generally, habitat radiomic models outperformed traditional radiomic models in stratifying prognosis for HCC. The habitat radiomic model based on the segmented habitat 4 involving decision tree (DT) screening and random forest (RF) classifier was identified as the optimal model with an AUC_mean of 0.806. Distinct resting natural killer (NK) cell infiltrations significantly contributed to the prognosis stratification of HCC by the optimal habitat radiomic model. The habitat radiomic model based on CT images was potentially predictive of overall survival for HCC, with a superiority over the traditional radiomic model. The prognostic power of the habitat radiomic model was partly attributed to the distinct immune status captured in the CT images.

To compare the performance of large language model (LLM) based Gemini and Generative Pre-trained Transformers (GPTs) in data mining and generating structured reports based on free-text PET/CT reports for breast cancer after user-defined tasks. Breast cancer patients (mean age, 50 years ± 11 [SD]; all female) who underwent consecutive ¹⁸F-FDG PET/CT for follow-up between July 2005 and October 2023 were retrospectively included in the study. A total of twenty reports from 10 patients were used to train user-defined text prompts for Gemini and GPTs, by which structured PET/CT reports were generated. The natural language processing (NLP) generated structured reports and the structured reports annotated by nuclear medicine physicians were compared in terms of data extraction accuracy and capacity of progress decision-making. Statistical methods, including chi-square test, McNemar test and paired samples t-test, were employed in the study. The structured PET/CT reports for 131 patients were generated by using the two NLP techniques, including Gemini and GPTs. In general, GPTs exhibited superiority over Gemini in data mining in terms of primary lesion size (89.6% vs. 53.8%, p < 0.001) and metastatic lesions (96.3% vs 89.6%, p < 0.001). Moreover, GPTs outperformed Gemini in making decision for progress (p < 0.001) and semantic similarity (F1 score 0.930 vs 0.907, p < 0.001) for reports. GPTs outperformed Gemini in generating structured reports based on free-text PET/CT reports, which is potentially applied in clinical practice. The data used and/or analyzed during the current study are available from the corresponding author on reasonable request.

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

Backgrounds: To evaluate the predictive power of 18F-Fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) derived radiomics in molecular subtype classification of breast cancer (BC). Methods: A total of 273 primary BC patients who underwent a 18F-FDG PET/CT imaging prior to any treatment were included in this retrospective study, and the values of five conventional PET parameters were calculated, including the maximum standardized uptake value (SUVmax), SUVmean, SUVpeak, metabolic tumor volume (MTV), and total lesion glycolysis (TLG). The ImageJ 1.50i software and METLAB package were used to delineate the contour of BC lesions and extract PET/CT derived radiomic features reflecting heterogeneity. Then, the least absolute shrinkage and selection operator (LASSO) algorithm was used to select optimal subsets of radiomic features and establish several corresponding radiomic signature models. The predictive powers of individual PET parameters and developed PET/CT derived radiomic signature models in molecular subtype classification of BC were evaluated by using receiver operating curves (ROCs) analyses with areas under the curve (AUCs) as the main outcomes. Results: All of the three SUV parameters but not MTV nor TLG were found to be significantly underrepresented in luminal and non-triple (TN) subgroups in comparison with corresponding non-luminal and TN subgroups. Whereas, no significant differences existed in all the five conventional PET parameters between human epidermal growth factor receptor 2+ (HER2+) and HER2– subgroups. Furthermore, all of the developed radiomic signature models correspondingly exhibited much more better performances than all the individual PET parameters in molecular subtype classification of BC, including luminal vs. non-luminal, HER2+ vs. HER2–, and TN vs. non-TN classification, with a mean value of 0.856, 0.818, and 0.888 for AUC. Conclusions: PET/CT derived radiomic signature models outperformed individual significant PET parameters in molecular subtype classification of BC.

Objective To investigate the value of ¹⁸F-FDG PET/CT-based intratumoral and peritumoral radiomics in predicting the efficacy of neoadjuvant chemotherapy (NAC) for breast cancer. Methods 190 patients who met the inclusion and exclusion criteria from 2017 to 2022 were studied. Features were extracted from the PET/CT intratumoral and peritumoral regions, feature selection was performed through the correlation analysis, t-tests, and least absolute shrinkage and selection operator regression (LASSO). Four classifiers, support vector machine (SVM), k-nearest neighbor (KNN), logistic regression (LR), and naive bayes (NB) were used to build the prediction models. The receiver operating characteristic (ROC) curves were plotted to measure the predictive performance of the models. Concurrent stratified analysis was conducted to establish subtype-specific features for each molecular subtype. Results Compared to intratumoral features alone, intratumoral + peritumoral features achieved higher AUC values in each classifier. The SVM model constructed with intratumoral + peritumoral features achieved the highest AUC values in both the train and test set (train set: 0.95 and test set: 0.83). Subtype-specific features improve performance in predicting the efficacy of NAC (luminal group: 0.90; HER2 + group: 0.86; triple negative group: 0.92). Conclusion Intratumoral and peritumoral radiomics models based on ¹⁸F-FDG PET/CT can reliably forecast the efficacy of NAC, thereby assisting clinical decision-making.