In recent decades, artificial intelligence (AI) tools have been applied in many medical fields, opening the possibility of finding novel solutions for managing very complex and multifactorial problems, such as those commonly encountered in radiotherapy (RT). We conducted a PubMed and Scopus search to identify the AI application field in RT limited to the last four years. In total, 1824 original papers were identified, and 921 were analyzed by considering the phase of the RT workflow according to the applied AI approaches. AI permits the processing of large quantities of information, data, and images stored in RT oncology information systems, a process that is not manageable for individuals or groups. AI allows the iterative application of complex tasks in large datasets (e.g., delineating normal tissues or finding optimal planning solutions) and might support the entire community working in the various sectors of RT, as summarized in this overview. AI-based tools are now on the roadmap for RT and have been applied to the entire workflow, mainly for segmentation, the generation of synthetic images, and outcome prediction. Several concerns were raised, including the need for harmonization while overcoming ethical, legal, and skill barriers.

Background: Current prognostic models lack the use of pre-operative CT images to predict recurrence in endometrial cancer (EC) patients. Our study aimed to investigate the potential of radiomic features extracted from pre-surgical CT scans to accurately predict disease-free survival (DFS) among EC patients. Methods: Contrast-Enhanced CT (CE-CT) scans from 81 EC cases were used to extract the radiomic features from semi-automatically contoured volumes of interest. We employed a 10-fold cross-validation approach with a 6:4 training to test set and utilized data augmentation and balancing techniques. Univariate analysis was applied for feature reduction leading to the development of three distinct machine learning (ML) models for the prediction of DFS: LASSO-Cox, CoxBoost and Random Forest (RFsrc). Results: In the training set, the ML models demonstrated AUCs ranging from 0.92 to 0.93, sensitivities from 0.96 to 1.00 and specificities from 0.77 to 0.89. In the test set, AUCs ranged from 0.86 to 0.90, sensitivities from 0.89 to 1.00 and specificities from 0.73 to 0.90. Patients classified as having a high recurrence risk prediction by ML models exhibited significantly worse DSF (p-value < 0.001) across all models. Conclusions: Our findings demonstrate the potential of radiomics in predicting EC recurrence. While further validation studies are needed, our results underscore the promising role of radiomics in forecasting EC outcomes.

Background/Objectives: The aim of this study was the early identification of endometriosis-associated ovarian cancer (EAOC) versus non-endometriosis associated ovarian cancer (NEOC) or non-cancerous tissues using pre-surgery contrast-enhanced-Computed Tomography (CE-CT) images in patients undergoing surgery for suspected ovarian cancer (OC). Methods: A prospective trial was designed to enroll patients undergoing surgery for suspected OC. Volumes of interest (VOIs) were semiautomatically segmented on CE-CT images and classified according to the histopathological results. The entire dataset was divided into training (70%), validation (10%), and testing (20%). A Python pipeline was developed using the transfer learning approach, adopting four different convolution neural networks (CNNs). Each architecture (i.e., VGG19, Xception, ResNet50, and DenseNet121) was trained on each of the axial slices of CE-CT images and refined using the validation dataset. The results of each CNN model for each slice within a VOI were combined using three rival machine learning (ML) models, i.e., Random Forest (RF), Gradient Boosting (GB), and K-Nearest Neighbor (KNN), to obtain a final output distinguishing between EAOC and NEOC, and between EAOC/NEOC and non-tumoral tissues. Furthermore, the performance of each hybrid model and the majority voting ensemble of the three competing ML models were evaluated using trained and refined hybrid CNN models combined with Support Vector Machine (SVM) algorithms, with the best-performing model selected as the benchmark. Each model’s performance was assessed based on the area under the receiver operating characteristic (ROC) curve (AUC), F1-score, sensitivity, and specificity. These metrics were then integrated into a Machine Learning Cumulative Performance Score (MLcps) to provide a comprehensive evaluation on the test dataset. Results: An MLcps value of 0.84 identified the VGG19 + majority voting ensemble as the optimal model for distinguishing EAOC from NEOC, achieving an AUC of 0.85 (95% CI: 0.70–0.98). In contrast, the VGG19 + SVM model, with an MLcps value of 0.76, yielded an AUC of 0.79 (95% CI: 0.63–0.93). For differentiating EAOC/NEOC from non-tumoral tissues, the VGG19 + SVM model demonstrated superior performance, with an MLcps value of 0.93 and an AUC of 0.97 (95% CI: 0.92–1.00). Conclusions: Hybrid models based on CE-CT have the potential to differentiate EAOC and NEOC patients as well as between OC (EAOC and NEOC) and non-tumoral ovaries, thus potentially supporting gynecological surgeons in personalized surgical approaches such as more conservative procedures.

Background: To evaluate the diagnostic performance of PSMA-PET compared to conventional imaging/liver biopsy in the detection of liver metastases in CRPC patients. Moreover, we evaluated a PSMA-PET/CT-based radiomic model able to identify liver metastases. Methods: Multicenter retrospective study enrolling patients with the following inclusion criteria: (a) proven CRPC patients, (b) PSMA-PET and conventional imaging/liver biopsy performed in a 6 months timeframe, (c) no therapy changes between PSMA-PET and conventional imaging/liver biopsy. PSMA-PET sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for liver metastases were calculated. After the extraction of radiomic features, a prediction model for liver metastases identification was developed. Results: Sixty CRPC patients were enrolled. Within 6 months before or after PSMA-PET, conventional imaging and liver biopsy identified 24/60 (40%) patients with liver metastases. PSMA-PET sensitivity, specificity, PPV, NPV, and accuracy for liver metastases were 0.58, 0.92, 0.82, 0.77, and 0.78, respectively. Either number of liver metastases and the maximum lesion diameter were significantly associated with the presence of a positive PSMA-PET (p < 0.05). On multivariate regression analysis, the radiomic feature-based model combining sphericity, and the moment of inverse difference (Idm), had an AUC of 0.807 (95% CI:0.686-0.920). Conclusion: For liver metastases assessment, [68Ga]Ga-PSMA-11-PET demonstrated moderate sensitivity while high specificity, PPV, and inter-reader agreement compared to conventional imaging/liver biopsy in CRPC patients.

Background: High‐dose rate brachytherapy using a non‐sealed ¹⁸⁸Rhenium resin (¹⁸⁸Re) is a recently approved treatment option for non‐melanoma skin cancer (NMSC). The treatment goal is to deliver a personalized absorbed dose to the deepest point of neoplastic infiltration corresponding to the minimal target dose. The treatment consists of the application of a ¹⁸⁸Re‐based resin over a plastic foil placed on the target skin surface. However, there is no treatment planning tool to assess the ¹⁸⁸Re activity needed for a personalized treatment.Purpose: The paper aims to present a novel Monte Carlo (MC)‐based tool for ¹⁸⁸Re‐based resin activity and dose calculation, experimentally validated using Gafchromic EBT3 films.Methods: MC simulations were carried out using FLUKA modeling density and composition of ¹⁸⁸Re resin. The MC‐based look up table (LUT) was incorporated in an ad hoc developed tool. The proposed tool allows the personalized calculation of treatment parameters (i.e., activity to be dispensed, the treatment duration, and dose volume histograms), according to the target dimension. The proposed tool was compared using Bland–Altman analysis to the previous calculation approaches conducted using VARSKIN in a retrospective cohort of 76 patients. The tool was validated in ad hoc experimental set ups using a stack of calibrated Gafchromic EBT3 films covered by a plastic film and exposed using a homogenous activity distribution of ¹⁸⁸Re eluate and a heterogeneous activity distribution of ¹⁸⁸Re resin mimic the patient treatment.Results: The agreement between the proposed tool and VARSKIN was evaluated on the investigated cohort with median range of target area, target depth, and treatment time equal to 4.8 [1.0–60.1] cm², 1.1 [0.2–3.0] mm, and 70 [21–285] min, with a median range of target dose (Gy) of 23.5 [10–54.9]. The calculated minimal target doses, ranged from 1% to 10% for intermediate target depths (1.2 ± 0.7 mm), while showing significant differences in the estimation of superficial (maximal) target doses. The agreement between MC calculation and measurements at different plans in a stack of Gafchromic EBT3 films was within 10% for both the homogenous and heterogeneous activity distribution of ¹⁸⁸Re. Worst agreements were observed for absorbed doses lower than 0.3 Gy.Conclusions: Our results support the implementation of our MC‐based tool in the practical routine for calculating the ¹⁸⁸Re resin activity and treatment parameters necessary for obtaining the prescribed minimal target dose.

Background: The accurate discrimination of uterine leiomyosarcomas and leiomyomas in a pre-operative setting remains a current challenge. To date, the diagnosis is made by a pathologist on the excised tumor. The aim of this study was to develop a machine learning algorithm using radiomic data extracted from contrast-enhanced computed tomography (CECT) images that could accurately distinguish leiomyosarcomas from leiomyomas. Methods: Pre-operative CECT images from patients submitted to surgery with a histological diagnosis of leiomyoma or leiomyosarcoma were used for the region of interest identification and radiomic feature extraction. Feature extraction was conducted using the PyRadiomics library, and three feature selection methods combined with the general linear model (GLM), random forest (RF), and support vector machine (SVM) classifiers were built, trained, and tested for the binary classification task (malignant vs. benign). In parallel, radiologists assessed the diagnosis with or without clinical data. Results: A total of 30 patients with leiomyosarcoma (mean age 59 years) and 35 patients with leiomyoma (mean age 48 years) were included in the study, comprising 30 and 51 lesions, respectively. Out of nine machine learning models, the three feature selection methods combined with the GLM and RF classifiers showed good performances, with predicted area under the curve (AUC), sensitivity, and specificity ranging from 0.78 to 0.97, from 0.78 to 1.00, and from 0.67 to 0.93, respectively, when compared to the results obtained from experienced radiologists when blinded to the clinical profile (AUC = 0.73 95%CI = 0.62–0.84), as well as when the clinical data were consulted (AUC = 0.75 95%CI = 0.65–0.85). Conclusions: CECT images integrated with radiomics have great potential in differentiating uterine leiomyomas from leiomyosarcomas. Such a tool can be used to mitigate the risks of eventual surgical spread in the case of leiomyosarcoma and allow for safer fertility-sparing treatment in patients with benign uterine lesions.