Purpose: The potential of large language models in medicine for education and decision-making purposes has been demonstrated as they have achieved decent scores on medical exams such as the United States Medical Licensing Exam (USMLE) and the MedQA exam. This work aims to evaluate the performance of ChatGPT-4 in the specialized field of radiation oncology.Methods: The 38th American College of Radiology (ACR) radiation oncology in-training (TXIT) exam and the 2022 Red Journal Gray Zone cases are used to benchmark the performance of ChatGPT-4. The TXIT exam contains 300 questions covering various topics of radiation oncology. The 2022 Gray Zone collection contains 15 complex clinical cases.Results: For the TXIT exam, ChatGPT-3.5 and ChatGPT-4 have achieved the scores of 62.05% and 78.77%, respectively, highlighting the advantage of the latest ChatGPT-4 model. Based on the TXIT exam, ChatGPT-4’s strong and weak areas in radiation oncology are identified to some extent. Specifically, ChatGPT-4 demonstrates better knowledge of statistics, CNS & eye, pediatrics, biology, and physics than knowledge of bone & soft tissue and gynecology, as per the ACR knowledge domain. Regarding clinical care paths, ChatGPT-4 performs better in diagnosis, prognosis, and toxicity than brachytherapy and dosimetry. It lacks proficiency in in-depth details of clinical trials. For the Gray Zone cases, ChatGPT-4 is able to suggest a personalized treatment approach to each case with high correctness and comprehensiveness. Importantly, it provides novel treatment aspects for many cases, which are not suggested by any human experts.Conclusion: Both evaluations demonstrate the potential of ChatGPT-4 in medical education for the general public and cancer patients, as well as the potential to aid clinical decision-making, while acknowledging its limitations in certain domains. Owing to the risk of hallucinations, it is essential to verify the content generated by models such as ChatGPT for accuracy.

Due to its superior soft tissue contrast, magnetic resonance imaging (MRI) is essential for many radiotherapy treatment indications. This is especially true for treatment planning in intracranial tumors, where MRI has a long-standing history for target delineation in clinical practice. Despite its routine use, care has to be taken when selecting and acquiring MRI studies for the purpose of radiotherapy treatment planning. Requirements on MRI are particularly demanding for intracranial stereotactic radiotherapy, where accurate imaging has a critical role in treatment success. However, MR images acquired for routine radiological assessment are frequently unsuitable for high-precision stereotactic radiotherapy as the requirements for imaging are significantly different for radiotherapy planning and diagnostic radiology. To assure that optimal imaging is used for treatment planning, the radiation oncologist needs proper knowledge of the most important requirements concerning the use of MRI in brain stereotactic radiotherapy. In the present review, we summarize and discuss the most relevant issues when using MR images for target volume delineation in intracranial stereotactic radiotherapy.

Background. In this study, we investigated the prognostic role of homotypic tumor cell cannibalism in different cancer types.Methods. The phenomenon of one cell being internalized into another, which we refer to as “cell-in-cell event,” was assessed in 416 cases from five head and neck cancer cohorts, as well as one anal and one rectal cancer cohort. The samples were processed into tissue microarrays and immunohistochemically stained for E-cadherin and cleaved caspase-3 to visualize cell membranes and apoptotic cell death.Results. Cell-in-cell events were found in all of the cohorts. The frequency ranged from 0.7 to 17.3 cell-in-cell events per mm². Hardly any apoptotic cells were found within the cell-in-cell structures, although apoptotic cell rates were about 1.6 to two times as high as cell-in-cell rates of the same tissue sample. High numbers of cell-in-cell events showed adverse effects on patients’ survival in the head and neck and in the rectal cancer cohorts. In multivariate analysis, high frequency was an adverse prognostic factor for overall survival in patients with head and neck cancer (p=0.008).Conclusion. Cell-in-cell events were found to predict patient outcomes in various types of cancer better than apoptosis and proliferation and might therefore be used to guide treatment strategies.

Background: While the role of stereotactic radiotherapy for brain metastases is increasing, evidence on the comparative efficacy and safety of fractionated stereotactic radiotherapy (FSRT) and single-session radiosurgery (SRS) is scarce. Methods: Longitudinal volumetric analysis was performed in a consecutive cohort of 120 patients and 190 brain metastases (>0.065 cm³ in volume / > ~5 mm in diameter) treated exclusively with FSRT (n = 98) and SRS (n = 92), respectively. A total of 972 tumor segmentations was used, averaging 5.1 time points per metastasis. Progression was defined using a volumetric extension of the RANO-BM criteria. Local control and radionecrosis were compared for lesions treated with FSRT and SRS, respectively. Results: Metastases treated with FSRT were significantly larger at baseline (mean, 4.66 vs. 0.40 cm³, p < 0.001). Biologically effective dose (BED) for metastases (α/β = 12, linear-quadratic-cubic model) was significantly associated with local control, whereas BED for normal brain (α/β = 2, linear-quadratic model) was significantly associated with radionecrosis. Median time to local progression was 22.9 months in the FSRT group compared to 14.5 months in the SRS group (p = 0.022). Overall radionecrosis rate at 12 months was 3.4% for FSRT and 14.8% for SRS (p = 0.010). Radionecrosis °IV requiring resection with histologic proof of radiation necrosis also was significantly reduced in the FSRT group (FSRT 0.0% vs. SRS 3.9%, p = 0.041). In multivariate analysis, FSRT was associated with reduced risk of progression (HR 0.47, p = 0.015) and reduced risk of radionecrosis (HR 0.18, p = 0.045). Conclusions: This volumetric study provides initial evidence that the improvements in therapeutic ratio expected for FSRT in larger brain metastases, might equally extend into the domain of smaller metastases, traditionally less considered for fractionated treatment. FSRT might constitute an important tool to further increase local control and reduce radionecrosis risk in stereotactic radiotherapy for brain metastases, that should be assessed in randomized intervention trials.

Background: Promptable foundation auto-segmentation models like Segment Anything (SA, Meta AI, New York, USA) represent a novel class of universal deep learning auto-segmentation models that could be employed for interactive tumor auto-contouring in RT treatment planning. Methods: Segment Anything was evaluated in an interactive point-to-mask auto-segmentation task for glioma brain tumor auto-contouring in 16,744 transverse slices from 369 MRI datasets (BraTS 2020 dataset). Up to nine interactive point prompts were automatically placed per slice. Tumor boundaries were auto-segmented on contrast-enhanced T1w sequences. Out of the three auto-contours predicted by SA, accuracy was evaluated for the contour with the highest calculated IoU (Intersection over Union, “oracle mask,” simulating interactive model use with selection of the best tumor contour) and for the tumor contour with the highest model confidence (“suggested mask”). Results: Mean best IoU (mbIoU) using the best predicted tumor contour (oracle mask) in full MRI slices was 0.762 (IQR 0.713–0.917). The best 2D mask was achieved after a mean of 6.6 interactive point prompts (IQR 5–9). Segmentation accuracy was significantly better for high- compared to low-grade glioma cases (mbIoU 0.789 vs. 0.668). Accuracy was worse using the suggested mask (0.572). Stacking best tumor segmentations from transverse MRI slices, mean 3D Dice score for tumor auto-contouring was 0.872, which was improved to 0.919 by combining axial, sagittal, and coronal contours. Conclusion: The Segment Anything foundation segmentation model can achieve high accuracy for glioma brain tumor segmentation in MRI datasets. The results suggest that foundation segmentation models could facilitate RT treatment planning when properly integrated in a clinical application.