Exploring Capabilities of Large Language Models such as ChatGPT in Radiation Oncology.

Purpose: Technological progress of machine learning and natural language processing has led to the development of large language models (LLMs), capable of producing well-formed text responses and providing natural language access to knowledge. Modern conversational LLMs such as ChatGPT have shown remarkable capabilities across a variety of fields, including medicine. These models may assess even highly specialized medical knowledge within specific disciplines, such as radiation therapy. We conducted an exploratory study to examine the capabilities of ChatGPT to answer questions in radiation therapy. Methods and Materials: A set of multiple-choice questions about clinical, physics, and biology general knowledge in radiation oncology as well as a set of open-ended questions were created. These were given as prompts to the LLM ChatGPT, and the answers were collected and analyzed. For the multiple-choice questions, it was checked how many of the answers of the model could be clearly assigned to one of the allowed multiple-choice-answers, and the proportion of correct answers was determined. For the open-ended questions, independent blinded radiation oncologists evaluated the quality of the answers regarding correctness and usefulness on a 5-point Likert scale. Furthermore, the evaluators were asked to provide suggestions for improving the quality of the answers. Results: For 70 multiple-choice questions, ChatGPT gave valid answers in 66 cases (94.3%). In 60.61% of the valid answers, the selected answer was correct (50.0% of clinical questions, 78.6% of physics questions, and 58.3% of biology questions). For 25 open-ended questions, 12 answers of ChatGPT were considered as "acceptable," "good," or "very good" regarding both correctness and helpfulness by all 6 participating radiation oncologists. Overall, the answers were considered "very good" in 29.3% and 28%, "good" in 28% and 29.3%, "acceptable" in 19.3% and 19.3%, "bad" in 9.3% and 9.3%, and "very bad" in 14% and 14% regarding correctness/helpfulness. Conclusions: Modern conversational LLMs such as ChatGPT can provide satisfying answers to many relevant questions in radiation therapy. As they still fall short of consistently providing correct information, it is problematic to use them for obtaining medical information. As LLMs will further improve in the future, they are expected to have an increasing impact not only on general society, but also on clinical practice, including radiation oncology.

Applications of Machine Learning in Palliative Care: A Systematic Review.

Objective: To summarize the available literature on using machine learning (ML) for palliative care practice as well as research and to assess the adherence of the published studies to the most important ML best practices. Methods: The MEDLINE database was searched for the use of ML in palliative care practice or research, and the records were screened according to PRISMA guidelines. Results: In total, 22 publications using machine learning for mortality prediction (n = 15), data annotation (n = 5), predicting morbidity under palliative therapy (n = 1), and predicting response to palliative therapy (n = 1) were included. Publications used a variety of supervised or unsupervised models, but mostly tree-based classifiers and neural networks. Two publications had code uploaded to a public repository, and one publication uploaded the dataset. Conclusions: Machine learning in palliative care is mainly used to predict mortality. Similarly to other applications of ML, external test sets and prospective validations are the exception.

Radiotherapy for glioblastoma patients with poor performance status.

PURPOSE: There is limited information on treatment recommendations for glioblastoma patients with poor performance status. Here, we aim to evaluate the association of radiotherapy on survival in glioblastoma patients presenting with poor postoperative performance status in first-line setting. METHODS: We retrospectively analyzed data of 93 glioblastoma patients presenting with poor postoperative performance status (ECOG 2-4) at the University Hospital Zurich, Switzerland, in the years 2005-2019. A total of 43 patients received radiotherapy with or without systemic therapy in the first-line setting, whereas 50 patients received no additive local or systemic treatment after initial biopsy or resection. Overall survival was calculated from primary diagnosis and from the end of radiotherapy. In addition, factors influencing survival were analyzed. RESULTS: Median overall survival from primary diagnosis was 6.2 months in the radiotherapy group (95% CI 6.2-14.8 weeks, range 2-149 weeks) and 2.3 months in the group without additive treatment (95% CI 1.3-7.4 weeks, range 0-28 weeks) (p < 0.001). This survival benefit was confirmed by landmark analyses. Factors associated with overall survival were extent of resection and administration of radiotherapy with or without systemic treatment. Median survival from end of radiotherapy was 3 months (95% CI 4.3-21.7 weeks, range 0-72 weeks), with 25.6% (n = 11) early termination of treatment and 83.7% (n = 36) requiring radiotherapy as in-patients. Performance status improved in 27.9% (n = 12) of patients after radiotherapy. CONCLUSION: In this retrospective single-institution analysis, radiotherapy improved overall survival in patients with poor performance status, especially in patients who were amendable to neurosurgical resection.

Stereotactic Radiotherapy after Radical Prostatectomy in Patients with Prostate Cancer in the Adjuvant or Salvage Setting: A Systematic Review.

(1) Background: Prostate cancer is the most common cancer in men and can be treated with radical prostatectomy (RPE) or radiotherapy in the primary setting. Stereotactic radiotherapy (SBRT) has proven to be effective and well tolerated in this setting. However, if SBRT is an equally promising treatment option if applied in the adjuvant or salvage setting after RPE remains unknown. (2) Methods: We searched the PubMed and Embase databases with the following full-text queries in August 2021 for any combination of the terms "SBRT", "prostate", "adjuvant", "postoperative", "salvage", "stereotactic radiotherapy", "prostate bed". There were no limitations regarding publication date or language. We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations. (3) Results: We identified 11 individual studies that were included in this systematic review. Three publications included patients without prior radiotherapy and the remaining eight patients with prior radiotherapy. In all but two publications the radiation target was the macroscopic recurrence. SBRT was overall well tolerated with acceptable rates of acute and late gastrointestinal or genitourinary toxicity. Quality of life was published for two phase I trials with good results. There was a very heterogeneous reporting on biochemical control after SBRT. (4) Conclusions: At this point, ultra-hypofractionated RT using SBRT to the prostate bed remains experimental and its use should be restricted to clinical trials. Given the biological rationale for extreme hypofractionation in patients with prostate cancer and the acceptable toxicity rates that have been reported, further exploration of this field is warranted.

Convolutional Neural Networks to Detect Vestibular Schwannomas on Single MRI Slices: A Feasibility Study.

In this study. we aimed to detect vestibular schwannomas (VSs) in individual magnetic resonance imaging (MRI) slices by using a 2D-CNN. A pretrained CNN (ResNet-34) was retrained and internally validated using contrast-enhanced T1-weighted (T1c) MRI slices from one institution. In a second step, the model was externally validated using T1c- and T1-weighted (T1) slices from a different institution. As a substitute, bisected slices were used with and without tumors originating from whole transversal slices that contained part of the unilateral VS. The model predictions were assessed based on the categorical accuracy and confusion matrices. A total of 539, 94, and 74 patients were included for training, internal validation, and external T1c validation, respectively. This resulted in an accuracy of 0.949 (95% CI 0.935-0.963) for the internal validation and 0.912 (95% CI 0.866-0.958) for the external T1c validation. We suggest that 2D-CNNs might be a promising alternative to 2.5-/3D-CNNs for certain tasks thanks to the decreased demand for computational power and the fact that there is no need for segmentations. However, further research is needed on the difference between 2D-CNNs and more complex architectures.

Radiotherapy or Surgery? Comparative, Qualitative Assessment of Online Patient Education Materials on Prostate Cancer.

As multiple different treatment options are available for prostate cancer (PCa) and YouTube is commonly used as a source for medical information, we performed a systematic and comparative assessment of available videos guiding patients on their choice for the optimal treatment. An independent search for surgical therapy or radiotherapy of PCa on YouTube was performed and the 40 most viewed videos of both groups were analyzed. The validated DISCERN questionnaire and PEMAT were utilized to evaluate their quality and misinformation. The median overall quality of the videos was found to be low for surgery videos, while radiotherapy videos results reached a moderate quality. The median PEMAT understandability score was 60% (range 0-100%) for radiotherapy and 75% (range 40-100) for surgery videos. The radiotherapy videos contained less misinformation and were judged to be of higher quality. Summarized, the majority of the provided videos offer insufficient quality of content and are potentially subject to commercial bias without reports on possible conflict of interest. Thus, most of available videos on YouTube informing PCa patients about possible treatment methods are not suited for a balanced patient education or as a basis for the patient's decision.

Cochlea sparing optimized radiotherapy for nasopharyngeal carcinoma.

BACKGROUND: Definitive chemoradiotherapy (CRT) is standard of care for nasopharyngeal carcinoma. Due to the tumor localization and concomitant platinum-based chemotherapy, hearing impairment is a frequent complication, without defined dose-threshold. In this study, we aimed to achieve the maximum possible cochleae sparing. MATERIALS AND METHODS: Treatment plans of 20 patients, treated with CRT (6 IMRT and 14 VMAT) based on the QUANTEC organs-at-risk constraints were investigated. The cochleae were re-delineated independently by two radiation oncologists, whereas target volumes and other organs at risk (OARs) were not changed. The initial plans, aiming to a mean cochlea dose < 45 Gy, were re-optimized with VMAT, using 2-2.5 arcs without compromising the dose coverage of the target volume. Mean cochlea dose, PTV coverage, Homogeneity Index, Conformity Index and dose to other OAR were compared to the reference plans. Wilcoxon signed-rank test was used to evaluate differences, a p value < 0.05 was considered significant. RESULTS: The re-optimized plans achieved a statistically significant lower dose for both cochleae (median dose for left and right 14.97 Gy and 18.47 Gy vs. 24.09 Gy and 26.05 Gy respectively, p < 0.001) compared to the reference plans, without compromising other plan quality parameters. The median NTCP for tinnitus of the most exposed sites was 11.3% (range 3.52-91.1%) for the original plans, compared to 4.60% (range 1.46-90.1%) for the re-optimized plans (p < 0.001). For hearing loss, the median NTCP of the most exposed sites could be improved from 0.03% (range 0-99.0%) to 0.00% (range 0-98.5%, p < 0.001). CONCLUSIONS: A significantly improved cochlea sparing beyond current QUANTEC constraints is feasible without compromising the PTV dose coverage in nasopharyngeal carcinoma patients treated with VMAT. As there appears to be no threshold for hearing toxicity after CRT, this should be considered for future treatment planning.

Nationwide Survey of German Outpatient Gynecologic Oncology Practices during the Coronavirus Disease 2019 Pandemic: Reactions to the First Wave and Future Perspectives.

INTRODUCTION: In the spring of 2020, coronavirus disease 2019 posed a substantial challenge for countries and their healthcare systems. In Germany, over 70% of all cancer patients are treated in an outpatient setting, so gynecologic oncology practices are the guarantors of optimal patient care. We developed a survey to evaluate the management of gynecologic oncology patients. METHODS: The survey consisted of 38 questions and was sent to the members of the Berufsverband Niedergelassener Gynäkologischer Onkologen in Deutschland e.V. (BNGO), a professional association of gynecologic oncologists in the outpatient sector in Germany. RESULTS: The survey was completed by 54 out of 133 (41%) gynecologic oncologists from 14 out of 15 (93%) federal states where the BNGO is represented. Facing the pandemic, popular measures were mask requirements (100%), restriction of access to practices (94%), increased number of disinfectant dispensers (85%), installment of panes of acrylic glass (76%), or spatial alterations (67%). For most patients the pandemic had no influence on prioritization of therapies (82%) or prescribed systemic treatments (87%). Despite an increase in perceived psychological burden among the staff (72%), 85% (45/54) of the practices were not offered any additional psychological support. DISCUSSION AND CONCLUSION: As most cancer patients in Germany are treated in an outpatient setting, a suitable reaction of oncology centers to the new circumstances was crucial to secure optimal treatment and patient care. Nevertheless, the low prioritization of mental health or distress of healthcare workers poses a serious threat to the maintenance of optimal medical care in further waves of the pandemic.

Stereotactic Body Radiotherapy for High-Risk Prostate Cancer: A Systematic Review.

BACKGROUND: Radiotherapy (RT) is an established, potentially curative treatment option for all risk constellations of localized prostate cancer (PCA). Androgen deprivation therapy (ADT) and dose-escalated RT can further improve outcome in high-risk (HR) PCA. In recent years, shorter RT schedules based on hypofractionated RT have shown equal outcome. Stereotactic body radiotherapy (SBRT) is a highly conformal RT technique enabling ultra-hypofractionation which has been shown to be safe and efficient in patients with low- and intermediate-risk PCA. There is a paucity of data on the role of SBRT in HR PCA. In particular, the need for pelvic elective nodal irradiation (ENI) needs to be addressed. Therefore, we conducted a systematic review to analyze the available data on observed toxicities, ADT prescription practice, and oncological outcome to shed more light on the value of SBRT in HR PCA. METHODS: We searched the PubMed and Embase electronic databases for the terms "prostate cancer" AND "stereotactic" AND "radiotherapy" in June 2020. We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations. RESULTS: After a rigorous selection process, we identified 18 individual studies meeting all selection criteria for further analyses. Five additional studies were included because their content was judged as relevant. Three trials have reported on prostate SBRT including pelvic nodes; 2 with ENI and 1 with positive pelvic nodes only. The remaining studies investigated SBRT of the prostate only. Grade 2+ acute genitourinary (GU) toxicity was between 12% and 46.7% in the studies investigating pelvic nodes irradiation and ranged from 0% to 89% in the prostate only studies. Grade 2+ chronic GU toxicity was between 7% and 60% vs. 2% and 56.7%. Acute gastrointestinal (GI) grade 2+ toxicity was between 0% to 4% and 0% to 18% for studies with and without pelvic nodes irradiation, respectively. Chronic GI grade 2+ toxicity rates were between 4% and 50.1% vs. 0% and 40%. SBRT of prostate and positive pelvic nodes only showed similar toxicity rates as SBRT for the prostate only. Among the trials that reported on ADT use, the majority of HR PCA patients underwent ADT for at least 2 months; mostly neoadjuvant and concurrent. Biochemical control rates ranged from 82% to 100% after 2 years and 56% to 100% after 3 years. Only a few studies reported longer follow-up data. CONCLUSION: At this point, SBRT with or without pelvic ENI cannot be considered the standard of care in HR PCA, due to missing level 1 evidence. Treatment may be offered to selected patients at specialized centers with access to high-precision RT. While concomitant ADT is the current standard of care, the necessary duration of ADT in combination with SBRT remains unclear. Ideally, all eligible patients should be enrolled in clinical trials.

Convolutional Neural Networks for Classifying Laterality of Vestibular Schwannomas on Single MRI Slices-A Feasibility Study.

Introduction: Many proposed algorithms for tumor detection rely on 2.5/3D convolutional neural networks (CNNs) and the input of segmentations for training. The purpose of this study is therefore to assess the performance of tumor detection on single MRI slices containing vestibular schwannomas (VS) as a computationally inexpensive alternative that does not require the creation of segmentations. Methods: A total of 2992 T1-weighted contrast-enhanced axial slices containing VS from the MRIs of 633 patients were labeled according to tumor location, of which 2538 slices from 539 patients were used for training a CNN (ResNet-34) to classify them according to the side of the tumor as a surrogate for detection and 454 slices from 94 patients were used for internal validation. The model was then externally validated on contrast-enhanced and non-contrast-enhanced slices from a different institution. Categorical accuracy was noted, and the results of the predictions for the validation set are provided with confusion matrices. Results: The model achieved an accuracy of 0.928 (95% CI: 0.869-0.987) on contrast-enhanced slices and 0.795 (95% CI: 0.702-0.888) on non-contrast-enhanced slices from the external validation cohorts. The implementation of Gradient-weighted Class Activation Mapping (Grad-CAM) revealed that the focus of the model was not limited to the contrast-enhancing tumor but to a larger area of the cerebellum and the cerebellopontine angle. Conclusions: Single-slice predictions might constitute a computationally inexpensive alternative to training 2.5/3D-CNNs for certain detection tasks in medical imaging even without the use of segmentations. Head-to-head comparisons between 2D and more sophisticated architectures could help to determine the difference in accuracy, especially for more difficult tasks.

Impact of Low-Dose Irradiation of the Lung and Heart on Toxicity and Pulmonary Function Parameters after Thoracic Radiotherapy.

OBJECTIVE: To assess the impact of (low) dose irradiation to the lungs and heart on the incidence of pneumonitis and pulmonary function changes after thoracic radiotherapy (RT). Methods/Material: Data of 62 patients treated with curative thoracic radiotherapy were analyzed. Toxicity data and pulmonary function tests (PFTs) were obtained before RT and at 6 weeks, at 12 weeks, and at 6 months after RT. PFTs included ventilation (e.g., vital capacity) and diffusion parameters (e.g., diffusion capacity for carbon monoxide (DLCO)). Dosimetric data of the lung and heart were extracted to assess the impact of dose on PFT changes and radiation pneumonitis (RP). RESULTS: No statistically significant correlations between dose parameters and changes in ventilation parameters were found. There were statistically significant correlations between DLCO and low-dose parameters of the lungs (V5Gy-V30Gy (%)) and irradiation of the heart during the follow-up up to 6 months after RT, as well as a temporary correlation of the V60Gy (%) on the blood gas parameters at 12 weeks after RT. On multivariate analysis, both heart and lung parameters had a significant impact on DLCO. There was no statistically significant influence of any patient or treatment-related (including dose parameters) factors on the incidence of ≥G2 pneumonitis. CONCLUSION: There seems to be a lasting impact of low dose irradiation to the lung as well as irradiation to the heart on the DLCO after thoracic radiotherapy. No influence on RP was found in this analysis.