Clinical Radiation Oncology Research Articles

The continuous improvement of digital assistance in the radiation oncologist's work: from web-based nomograms to the adoption of large-language models (LLMs). A systematic review by the young group of the Italian association of radiotherapy and clinical oncology (AIRO).

Recently, the availability of online medical resources for radiation oncologists and trainees has significantly expanded, alongside the development of numerous artificial intelligence (AI)-based tools. This review evaluates the impact of web-based clinical decision-making tools in the clinical practice of radiation oncology. We searched databases, including PubMed, EMBASE, and Scopus, using keywords related to web-based clinical decision-making tools and radiation oncology, adhering to PRISMA guidelines. Out of 2161 identified manuscripts, 70 were ultimately included in our study. These papers all supported the evidence that web-based tools can be transversally integrated into multiple radiation oncology fields, with online applications available for dose and clinical calculations, staging and other multipurpose intents. Specifically, the possible benefit of web-based nomograms for educational purposes was investigated in 35 of the evaluated manuscripts. As regards to the applications of digital and AI-based tools to treatment planning, diagnosis, treatment strategy selection and follow-up adoption, a total of 35 articles were selected. More specifically, 19 articles investigated the role of these tools in heterogeneous cancer types, while nine and seven articles were related to breast and head & neck cancers, respectively. Our analysis suggests that employing web-based and AI tools offers promising potential to enhance the personalization of cancer treatment.

Large language foundation models encode clinical radiation oncology domain knowledge: Performance on the American College of Radiology Standardized Examination.

e13585 Background: The integration of Large Language Models (LLMs) into healthcare and medical education will represent a significant paradigm shift, offering transformative potential in how medical knowledge is accessed and assimilated. These models, however, have not yet been systematically trained, tested, or validated on complex medical information such as sub-specialty medical examinations. This study explores the performance of seven major LLMs in clinical radiation oncology using residency in-training exams. Methods: In this study, the 2021 American College of Radiology (ACR) Radiation Oncology In-Training Exam (TXIT) was used to evaluate the performance of various LLMs, including OpenAI's GPT-3.5-turbo, GPT-4, GPT-4-turbo, three Meta's Llama-2 models (7 billion, 13 billion, and 70 billion parameter), and Google's PaLM-2-text-bison. The ACR provided the publicly available national scoring for this exam. The exam comprised 298 questions across 13 domains, including clinical radiation oncology (195 questions, 65.4%). The exam was processed through each LLM via an application programming interface. LLM-generated answers were analyzed by clinical disease sites and compared to Radiation Oncology trainee performance and stratified by Post-Graduate Year (PGY) 2-5. Results: LLMs showed varied performance in the overall clinical radiation oncology domain, with OpenAI's GPT-4-turbo having the best performance with 68.0% correct answers, GPT-4 61.0%, GPT-3.5-turbo 48.0%, PaLM-2-text-bison 40.0%, and then the three Llama-2 models (70b 37.0%, 13b 38%, 7b 26%). GPT-4-turbo performed superiorly to lower-level (PGY2 51.6%, PGY3 61.6%) and comparably to upper-level radiation oncology trainees (PGY4 64.1%, PGY5 68.3%). Notably, GPT-4-turbo demonstrated 7.0% improvement over its predecessor GPT-4. LLMs scored the lowest in the gastrointestinal, genitourinary, and gynecology domains and highest in the bone and soft tissue, central nervous system and eye, and head, neck, and skin domains. Conclusions: GPT-4-turbo demonstrates clinical accuracy comparable to upper-level and superior to lower-level trainees in nearly all clinical domains. Conversely, Llama2 foundation models demonstrate overall worse performance than Level 1 (PGY2) trainees. Score discrepancies across disease site domains may be due to data availability, complexity of medical conditions, quality and quantity of training datasets, and interdisciplinary data inputs. Future research will need to evaluate the performance of models that are fine-tune trained in clinical oncology. This study also underscores the need for rigorous validation of LLM-generated information against established medical literature and expert consensus, necessitating expert oversight in their application in medical education and practice.

ACPSEM position paper: pre-treatment patient specific plan checks and quality assurance in radiation oncology.

The Australasian College of Physical Scientists and Engineers in Medicine (ACPSEM) has not previously made recommendations outlining the requirements for physics plan checks in Australia and New Zealand. A recent workforce modelling exercise, undertaken by the ACPSEM, revealed that the workload of a clinical radiation oncology medical physicist can comprise of up to 50% patient specific quality assurance activities. Therefore, in 2022 the ACPSEM Radiation Oncology Specialty Group (ROSG) set up a working group to address this issue. This position paper authored by ROSG endorses the recommendations of the American Association of Physicists in Medicine (AAPM) Task Group 218, 219 and 275 reports with some contextualisation for the Australia and New Zealand settings. A few recommendations from other sources are also endorsed to complete the position.

Prospective deployment of an automated implementation solution for artificial intelligence translation to clinical radiation oncology.

Artificial intelligence (AI)-based technologies embody countless solutions in radiation oncology, yet translation of AI-assisted software tools to actual clinical environments remains unrealized. We present the Deep Learning On-Demand Assistant (DL-ODA), a fully automated, end-to-end clinical platform that enables AI interventions for any disease site featuring an automated model-training pipeline, auto-segmentations, and QA reporting. We developed, tested, and prospectively deployed the DL-ODA system at a large university affiliated hospital center. Medical professionals activate the DL-ODA via two pathways (1): On-Demand, used for immediate AI decision support for a patient-specific treatment plan, and (2) Ambient, in which QA is provided for all daily radiotherapy (RT) plans by comparing DL segmentations with manual delineations and calculating the dosimetric impact. To demonstrate the implementation of a new anatomy segmentation, we used the model-training pipeline to generate a breast segmentation model based on a large clinical dataset. Additionally, the contour QA functionality of existing models was assessed using a retrospective cohort of 3,399 lung and 885 spine RT cases. Ambient QA was performed for various disease sites including spine RT and heart for dosimetric sparing. Successful training of the breast model was completed in less than a day and resulted in clinically viable whole breast contours. For the retrospective analysis, we evaluated manual-versus-AI similarity for the ten most common structures. The DL-ODA detected high similarities in heart, lung, liver, and kidney delineations but lower for esophagus, trachea, stomach, and small bowel due largely to incomplete manual contouring. The deployed Ambient QAs for heart and spine sites have prospectively processed over 2,500 cases and 230 cases over 9 months and 5 months, respectively, automatically alerting the RT personnel. The DL-ODA capabilities in providing universal AI interventions were demonstrated for On-Demand contour QA, DL segmentations, and automated model training, and confirmed successful integration of the system into a large academic radiotherapy department. The novelty of deploying the DL-ODA as a multi-modal, fully automated end-to-end AI clinical implementation solution marks a significant step towards a generalizable framework that leverages AI to improve the efficiency and reliability of RT systems.

THE ROLE OF REPAIR PROCESSES IN THE FORMATION OF RADIOSENSITIVITY OF HEALTHY AND TUMOUR CELLS IN HUMANS

Despite significant progress in the development and implementation of new radiotherapeutic techniques over the past decade and positive results of radiotherapy for cancer patients, overall and progression-free survival rates are still poor. The analysis of literature data and our own research in the field of clinical radiobiology and radiation oncology shows that cellular DNA repair systems increase the radioresistance of tumours and thus hinder the improvement of the effectiveness of patient therapy. The role of repair processes in shaping the radiosensitivity of tumours compared to healthy tissues is discussed in detail, and their differences are described. It is emphasized that the repair of irradiated DNA damage in tumour cells is more intensive than in healthy tissue cells due to increased expression of repair enzymes. Together with the instability of the tumour cell genome, this causes a large variation in their radiosensitivity and indicates the priority of research aimed at finding and developing radioprotectors to protect the genome of healthy cells from the environment of the irradiated tumour without affecting (reducing) its radiosensitivity. Such a way to optimize the results of therapeutic radiation will help minimize radiation complications that require additional therapy and worsen the quality of life of treated patients.

EXPERIENCE OF EXPERIMENTAL RESEARCH ON RADIATION ONCOLOGY AT R.E. KAVETSKY INSTITUTE OF EXPERIMENTAL PATHOLOGY, ONCOLOGY AND RADIOBIOLOGY, NAS OF UKRAINE

This article briefly summarizes our long-term experience of research in the field of experimental and clinical radiation oncology unified by the key word "radiosensitivity". Consistently presented and interpreted here are the main results on biodosimetry of irradiation depending on doses and quality of ionizing radiation and determination of individual radiosensitivity of cancer patients. We justified the use of radiomitigators to reduce the frequency and severity of post-radiation complications in cancer patients, and for radiation protection of the general population. The radioprotective effect of the antioxidant inosine in the somatic cells of cancer patients in the range of low radiation doses was demonstrated. We established that in persons who are hypersensitive to irradiation, the reparative potential is reduced by about 60% compared to ones with normal indices of individual radiosensitivity. Cytogenetic predictors of radiosensitivity of healthy cells adjacent to the irradiated tumor have been determined. Unfortunately, they have not yet become a point of application for individual planning of irradiation courses and assessment of severity of post-radiation complications. The intensive development of selective radioprotectors that would selectively protect healthy tissues in the course of radiation therapy, reducing their radiosensitivity by activating reparation processes, is considered a priority direction of modern radiation oncology.

MR-Linac Dosimetry: Current Approaches and Challenges

The use of MR-integrated linacs (MR-linac) in clinical radiation oncology applications are becoming more prevalent. However, given the novel and radically different designs of these systems from conventional linacs, current radiation dosimetry protocols for high energy photons are no longer appropriate for use in modern MR-linac systems: First, given the presence of the magnetic field and the linac design, traditional reference conditions defined by previous high energy photon dosimetry protocols cannot be met; Second, the presence of the strong magnetic field can affect the performance of conventional equipment used for dosimetry. In this manuscript, we describe some of the challenges faced in radiation dosimetry in external MR-guided radiotherapy delivery systems, summarize some of the publications in this area, and finally discuss the early work by the American Association of physicists in Medicine (AAPM) Task Group 351 which is mandated with producing a protocol for reference dosimetry in MR-linac units.

Current Strengths and Weaknesses of ChatGPT as a Resource for Radiation Oncology Patients and Providers

Purpose: ChatGPT, an artificial intelligence (AI) program that uses natural language processing to generate conversational-style responses to questions or inputs, is increasingly being used by both patients and healthcare professionals. This study aims to evaluate the accuracy and comprehensiveness of ChatGPT in radiation oncology-related domains, including answering common patient questions, summarizing landmark clinical research studies, and providing literature reviews with specific references supporting current standard of care clinical practice in radiation oncology. Methods: We assessed the performance of ChatGPT version 3.5 (ChatGPT3.5) in three areas. We evaluated ChatGPT3.5’s ability to answer 28 templated patient-centered questions applied across 9 cancer types. We then tested ChatGPT3.5’s ability to summarize specific portions of 10 landmark studies in radiation oncology. Next, we used ChatGPT3.5 to identify scientific studies supporting current standard of care practice in clinical radiation oncology for five different cancer types. Each response was graded independently by two reviewers, with discordant grades resolved by a third reviewer. Results: ChatGPT3.5 frequently generated inaccurate or incomplete responses. Only 39.7% of responses to patient-centered questions were considered correct and comprehensive. When summarizing landmark studies in radiation oncology, 35.0% of ChatGPT3.5′s responses were accurate and comprehensive, improving to 43.3% when provided the full text of the study. ChatGPT3.5′s ability to present a list of studies related to standard of care clinical practices was also unsatisfactory, with 50.6% of the provided studies fabricated. Conclusion: ChatGPT should not be considered a reliable radiation oncology resource for patients or providers at this time, as it frequently generates inaccurate or incomplete responses. However, natural language programming-based AI programs are rapidly evolving, and future versions of ChatGPT or similar programs may demonstrate improved performance in this domain.

Thymic epithelial tumour radiotherapy in the UK: A survey of current clinical practice

Significant variation in treatment centre setup and radiotherapy practice for thymic epithelial tumours (TET) was identified through a comprehensive survey of current UK Clinical (Radiation) Oncology practice. Multi-centre collaboration and wider TET specific multidisciplinary team meetings are needed and will be essential for developing expertise in TET radiotherapy.

Burnout Rates in Clinical Radiation Oncology and Variations across Professional Subgroups

Physician and medical professional burnout has emerged as an important issue in clinical medicine and has been shown to negatively affect patient outcomes. The aim of this study is to determine how rates of burnout vary between different radiation oncology professional subgroups. Participants were selected from those employed in one of the professional subgroups (radiation therapists, nurses, physicists, dosimetrists, and physicians) at two institutions that cover a total of seven sites. 126 individuals were recruited to participate. In addition to demographics, participants were asked to respond to questions regarding professional burnout utilizing the Maslach Burnout Inventory (MBI). Responses were on a seven-point Likert scale with subcategories of Exhaustion, Depersonalization, and Personal Achievement. Each participant's score per subcategory was categorized as either low, moderate, or high. For our primary outcome, frequencies and proportions were reported and Fisher's Exact Test was used to evaluate whether there was a statistically significant difference in proportions in categorized burnout scores across professional subgroups. Secondary analyses were performed comparing burnout across sex, age, and categorized years in position. An alpha level of 0.05 was used for all analyses. In total, 55 participants responded to the survey: 12 nurses, 8 dosimetrists, 12 physicians, 5 physicists, and 18 radiation therapists for a response rate of 43.6%. Demographics were consistent across professional subgroups with the exception of participant sex (p = .0397), which had a greater proportion of females (72.22%) in the Radiation Therapist subgroup. Among the three subcategories, only Exhaustion had a significant difference between groups (p = .0494) with Dosimetrists (62.50%) and Physicists (40.00%) experiencing the highest level of burnout. For Depersonalization, at least one individual in each subgroup identified high burnout with higher rates among the Dosimetrists (25.00%) and Physicists (40.00%). Similarly, at least one person in each subgroup reported high burnout in Personal Achievement, but the highest response was among Dosimetrists (37.50%). In total, at least 12.75% of participants reported high burnout in at least one subcategory. No statistically significant differences in burnout were detected in our secondary analyses across sex, age, and years in position. Our study showed that burnout rates are moderate in members of the radiation oncology team, with dosimetrists and physicists displaying the highest levels of burnout. In order to better estimate professional burnout, larger studies should be conducted to determine causes of burnout. Furthermore, wellness programs should be encouraged and incentivized to help reduce burnout and improve workplace morale, occupational fulfillment, and effective patient care.

A Survey of Submission and Review Standards across Clinical Radiation Oncology and Medical Physics Journals: Devising the Optimal Criteria for Reproducibility and Sustained Impact

Publishing and editorial policies differ substantially across the Radiation Oncology (RO) and Medical Physics (MedPhys) compendium of journals. Adoptance of modern standards in scientific publishing and data sharing have the potential to improve the impact and reliability of the RO literature. We characterized the editorial, authorship and peer reviewer policies of various prominent clinical RO (N = 16) and medical physics (N = 9) peer-reviewed journals affiliated with professional societies for characteristics that are associated with improved reproducibility and rigorous review. A combination of tools including Enhancing the QUAlity and Transparency Of health Research (EQUATOR), Findability, Accessibility, Interoperability, and Reuse (FAIR), and Quality Output Checklist and Content Assessment (QuOCCA) principles were used to quantify the value and reproducibility of journal policies. Cohen's kappa coefficient was utilized to assess agreement between reviewers. Components of the above tools were regressed against various scientometric indices (H-index, IF, etc.) to identify factors that are associated with perceived relative importance within the field. Reviewer agreement (κ) for scientometric indices was highest (1.0) for criteria for statistical review and data submission standards and lowest (-0.246) for various submission checklists. Data availability statements were endorsed (44%) or required (31%) in a higher proportion of RO journals relative to MedPhys journals (44%, 0% respectively). Data repository submission was required in <10% of journals. FAIR adoptance was poor (31%, 22%) in RO and MedPhys journals. ≥1 EQUATOR guideline checklist was endorsed or required in 76% of journals. While there were no glaring differences in editorial policies between RO and MedPhys journals, there was substantial heterogeneity of scientometrics evaluating the rigor of data submission, reproducibility standards, and statistical review criteria. Linear regression of journal impact factors indicated a predictive relationship between FAIR adoption standards, use of EQUATOR checklists, and more rigorous statistical method submission criteria. The present review documented and confirmed significant variation in submission, review, and publication policies across RO and MedPhys journals. Established scientometric standards, FAIR principle adoptance, and more rigorous statistical methodology were predictive of increasing journal impact factor.

Interpretable artificial intelligence in radiology and radiation oncology.

Artificial intelligence has been introduced to clinical practice, especially radiology and radiation oncology, from image segmentation, diagnosis, treatment planning and prognosis. It is not only crucial to have an accurate artificial intelligence model, but also to understand the internal logic and gain the trust of the experts. This review is intended to provide some insights into core concepts of the interpretability, the state-of-the-art methods for understanding the machine learning models, the evaluation of these methods, identifying some challenges and limits of them, and gives some examples of medical applications.

Early introduction to clinical radiation oncology through a self-directed competency-based medical student curriculum.

11038 Background: Specialty training is not traditionally incorporated into standard medical education and there is a deficiency in exposure to Radiation Oncology for “last minute” interested candidates, particularly from underrepresented backgrounds. Therefore, we implemented MERAO-200, an introductory 2-week in-person and virtual medical education curriculum without clinical prerequisites to facilitate early introduction to the field of Radiation Oncology and report its effect on student comfort and familiarity with radiation planning and contouring. Methods: Clinical students were enrolled in a 2-week elective without clinical prerequisites that integrated both clinical activities and self-directed learning through virtual contouring modules. The virtual curriculum was designed around three modules available on EduCase. A survey was conducted after completion of the modules that solicited self-identifying information and to assess their efficacy in increasing understanding of basic contouring skills. Paired t-tests were used to compare pre- and post-elective assessment results. Results: Twenty-three total students have completed MERAO-200 from Fall 2020 to Spring 2023. Of those, 15 submitted the survey. Eight (53%) students did not answer self-identifying information such as gender, race, ethnicity, or country of origin. Four (27%) students identified as Asian and 3 (20%) as white. No students identified as Hispanic, Latino or Spanish origin. Seven (46%) students identified as having USA as country of origin. Seven (47%) students reported interest in radiation oncology, medical oncology or surgical oncology. Three (20%) students were MS2, 7 (47%) were MS3, and 5 (33%) were MS4. Overall, there was a significant increase in comfort level using radiation planning software (p = 1.1e-4) as well as ability to find relevant contouring references (p = 4.2e-5). Knowledge of the anatomy and contouring principles also improved for the three modules: eye and brain (p = 2.8e-7), high grade glioma (3.2e-6), and head and neck nodal stations (4.5e-6). CNS module difficulty on a scale of 1-10 had a median difficulty score of 7 (range 3-8) and head and neck module difficulty was a median of 7 (range 3-10). Conclusions: Implementation of MERAO-200 improved medical student exposure to the field of Radiation Oncology and improved medical student comfort and familiarity with radiation planning and across all disease sites covered. More outreach work needs to be done to facilitate inclusion of minority student populations and students from underrepresented backgrounds. Early introduction at the medical student level could be a pathway to improve disparity in the field of Radiation Oncology.

Investigation of sponge medium for efficient concurrent tumor treating fields and radiotherapy for glioblastomas.

Realizing precise therapy for glioblastomas (GBMs), a kind of high-frequency malignant brain tumor, is of great importance in improving the overall survival (OS) of patients. With relentless efforts made in the past few years, a sponge medium has been introduced into concurrent tumor treating fields (TTFields) and radiotherapy to enhance therapy efficacy for GBMs, and some progresses have been witnessed. However, the specific physical and chemical characteristics of the sponge that can be used for GBMs have not been reported as far as we know. Therefore, this study aims to develop a simple yet robust method to select a candidate sponge medium and verify its safety in advanced concurrent TTFields and radiotherapy for GBMs through interdisciplinary investigation among materials science, medical physics, and clinical radiation oncology. Significantly, latex-free polyurethane (PU) sponges with a Hounsfield unit (HU) value lower than -750, which exhibit almost no negative influence on planning computed tomography (CT) imaging and radiotherapy dosimetry, are demonstrated to be available for concurrent TTFields and radiotherapy for GBMs. Moreover, in clinical research, the achieved clear CT images, negligible scalp toxicity, lower residual positioning errors, and high compliant rate of 82% over the selected representative sponge sample corroborate the availability and safety of PU sponges in practical applications for GBM treatment.

Automated planning stage tracking and analysis through an integrated whiteboard system

Purpose:One of the common challenges in delivering complex healthcare procedures such as radiation oncology is the organization and sharing of information in ways that facilitate workflow and prevent treatment delays. Within the major vendors of Oncology Information Systems (OIS) is a lack of tools and displays to assist in task timing and workflow processes. To address this issue, we developed an electronic whiteboard integrated with a local OIS to track, record, and evaluate time frames associated with clinical radiation oncology treatment planning processes. Methods:We developed software using an R environment hosted on a local web-server at Seattle Cancer Care Alliance (SCCA) in 2017. The planning process was divided into stages, and time-stamped moves between planning stages were recorded automatically via Mosaiq (Elekta, Sweden) Quality Check Lists (QCLs). Whiteboard logs were merged with Mosaiq-extracted diagnostic factors and evaluated for significance. Interventional changes to task time expectations were evaluated for 6 months in 2021 and compared with 6 month periods in 2018 and 2019. Results:Whiteboard/Mosaiq data from the SCCA show that treatment intent, number of prescriptions, and nodal involvement were main factors influencing overall time to plan completion. Contouring and Planning times were improved by 2.6 days (p<10−14) and 2.5 days (p<10−11), respectively. Overall time to plan completion was reduced by 33% (5.1 days; p<10−11). Conclusions:This report establishes the utility of real-time task tracking tools in a radiotherapy planning process. The whiteboard results provide data-driven evidence to add justification for practice change implementations.

ASTRO's Framework for Radiopharmaceutical Therapy Curriculum Development for Trainees

In 2017, the American Society for Radiation Oncology (ASTRO) board of directors prioritized radiopharmaceutical therapy (RPT) as a leading area for new therapeutic development, and the ASTRO RPT workgroup was created. Herein, the workgroup has developed a framework for RPT curriculum development upon which education leaders can build to integrate this modality into radiation oncology resident education. Through this effort, the workgroup aims to provide a guide to ensure robust training in an emerging therapeutic area within the context of existing radiation oncology training in radiation biology, medical physics, and clinical radiation oncology. The framework first determines the core RPT knowledge required to select patients, prescribe, safely administer, and manage related adverse events. Then, it defines the most important topics for preparing residents for clinical RPT planning and delivery. This framework is designed as a tool to supplement the current training that exists for radiation oncology residents. The final document was approved by the ASTRO board of directors in the fall of 2021.

Education in Radiation Oncology-Current Challenges and Difficulties.

The evolution and development of radiotherapy in the last two decades has meant that postgraduate medical training has not kept up with this rapid progress both in terms of multidisciplinary clinical approaches and especially in terms of technological advances. Education in radiation oncology is a major priority in the context of the rapid development of radiotherapy, including advanced knowledge of radiobiology, radiation physics and clinical oncology, anatomy, tumor biology and advanced medical imaging. In this context, the lack of training in radiation oncology in the curricula of medical faculties may have detrimental consequences for the training of residents in radiotherapy but also in their choice of specialty after completing their university studies. There is a clear gap between resident physicians’ actual and required knowledge of radiotherapy, and this requires urgent remediation. In the context of technical advances in imaging-guided radiotherapy (IGRT) and new radiobiology data, a balanced approach divided equally between general oncology, clinical radiation oncology, radiation oncology technology, medical physics and radiobiology, anatomy and multimodal imaging, including mentorship could bring educational and career choice benefits for students of radiation oncology.

Impact of the COVID-19 Pandemic on Radiation Oncology Consult Volumes Within a Multi-Site Academic Practice

Purpose/Objective(s)Cancer care has been significantly impacted by the COVID 19 pandemic; however, a comprehensive understanding of the downstream effects of the pandemic on radiation oncology practice in the United States has not been well quantified. We sought to quantify the differences in radiation consult volumes across varying disease sites and practice locations within our National Cancer Institute (NCI) designated Comprehensive Cancer Center, at time points preceding and during the COVID 19 pandemic. The purpose of this study is to identify potential gaps in care for specific disease sites that could addressed in the future.Materials/MethodsWe collected new patient referral data from three clinical radiation oncology sites for the 2019 calendar year and the 2020 calendar year. Overall new patient referrals for 2019 and 2020 were analyzed and compared as well as consult volume for each disease site.ResultsIn 2019, there was a total 4,284 overall consults and in 2020 there was a total of 3,442 consults. This is a 20% reduction in consults over a 12-month period. The month of April had the most significant decrease in consults with a total of 357 consults in April 2019 compared to a total of 179 consults in April 2020 (50% reduction). The month of February had the least reduction with 329 total consults in 2019 compared to a total of 313 consults in 2020 (5% reduction). There was a decrease in consult volumes for all tumor sites with the exception of following tumor sites which saw an increase in consult volumes: colon (55.6%), kidney &renal pelvis (15.8), larynx (32.7%), oral cavity (7%), Non-Hodgkin's Lymphoma (9.9%), other head & neck (13.9%), and thyroid cancer (36.4%). Breast and prostate cancer had the greatest absolute change in consult volume loss (32.6%, 29.3% respectively). Oral tongue cancer was the sole tumor site that had no change in consult volume across the study period.ConclusionThe ramifications of COVID-19 will be seen in the practices of clinicians, cancer centers, and the experiences of patients and their families for years to come. We found a 20% overall reduction in new patient referrals for 2020. The most significant absolute change in volume loss was seen in breast and prostate cancer. Head and neck cancer referrals were the most minimally impacted, with some increases in referrals during the study period which could possibly be related to limited surgical resources necessitating non-operative treatment. Further studies are needed to more comprehensively understand the impact of the COVID 19 pandemic on radiation oncology practice.

Radiation Oncology Provider Telehealth Satisfaction: Survey Results From a Single NCI-Designated Institution

Telehealth (TH) for radiation oncology services has emerged as a new modality for care delivery and will likely persist beyond the COVID pandemic. Data regarding radiation oncology provider satisfaction of TH are limited and essential to the sustainable adoption of this tool.An anonymous electronic survey assessing TH experience was distributed in 11/2020 to all clinical radiation oncology physicians and APPs at a single large NCI-designated institution, including affiliate clinics. Those who utilized TH (phone or video) were invited to participate. The provider survey was designed using the technology-acceptance model (TAM), a validated method to predict use and acceptance of technology tools. Survey items included 4 assessing provider role, 1 regarding TH utilization, 26 assessing TH experience on a 5-point Likert scale, and 1 free response assessing current barriers to TH. Percent satisfaction is reported as the percentage of top 2 positive or affirmative responses on the Likert scale as a proportion of all responses.19 of 34 radiation oncology providers (56%) completed the survey, including 15 attending physicians and 4 APPs. Providers specialized in central nervous system (n = 3), head and neck (n = 2), gastrointestinal (n = 1), breast (n = 2), genitourinary (n = 4), gynecological (n = 2), sarcoma (n = 1), and general oncology (n = 4). Providers reported having 1-10 (n = 5), 11-15 (n = 7), or > 20 years (n = 7) in practice. Providers utilized TH for on-treatment visits (53%), follow-ups (86%), and consults (79%). 56% of providers enjoyed experimenting with new technology and 61% felt that technological advances improved care for patients. Regarding aspects of the TH clinical encounter: providers had high satisfaction with ability to document the visit (89%), obtaining patient history (83%), and ease of discussing radiation treatment decisions (71%). There was lower satisfaction with ability to create rapport (33%), ease of obtaining consent for radiation (33%), and ease of evaluating physical exam findings (19%). Regarding workflow: 39% felt that TH was compatible with existing oncology clinical workflow, 39% felt TH gave them greater control over work, 33% providers felt that TH improved their job efficiency, and 28% felt TH made them more productive. Regarding ease of TH use: 44% felt that interacting with TH services was frustrating and 39% felt that TH services did not require much training. 24% felt the TH adequately replaced face to face visits. Providers identified the following barriers to TH implementation: lack of MA/RN/APP support, interruptions to TH visits by treatment/simulation clinical duties, lack of dedicated TH template, and burden of navigating the electronic medical record.Radiation oncology providers at our institution expressed mixed satisfaction to incorporating TH into their practice. Current strategies to address barriers, including implementation of a telehealth care coordinator, are underway.

Deep Learning Based Identification and Segmentation of Lung Tumors on Computed Tomography Images

Rapid and accurate estimation of tumor burden in biomedical images is essential for precisely monitoring cancer progression and assessing therapeutic response. The ability to detect and segment tumors using an automated approach is a key part of this task. Despite recent advances from deep learning, lung tumor delineation remains challenging, particularly when the tumor bounding box is not provided to the model. We hypothesized that clinical radiation oncology contours could supply a large enough dataset of 3D tumor segmentations to enable more accurate models. We developed and validated a deep learning-based model to identify and segment primary and metastatic lung tumors on computed tomography (CT) images.We curated a dataset consisting of CT images and clinical segmentations of 1,916 lung tumors in 1,504 patients who received radiation treatment for one or more primary or metastatic lung tumors. Segmentation quality was independently verified by a radiation oncologist using a custom web application. This dataset was used to train two 3D U-Net convolutional neural networks with varying model properties: one using high-resolution and small input volumes, and one using low-resolution and large input volumes. Models were ensembled together during validation. Performance was evaluated using an external held-out test set of CT images and segmentations from 59 patients with a single primary or metastatic lung tumor, treated at a separate clinical site. This test set consisted of 50 primary lung cancers and 9 metastases. To benchmark model performance against physicians, the test set was also contoured by two additional radiation oncologists.Median tumor volume in the external test set was 80.48 cubic centimeters (interquartile range [IQR]: 14.40 to 177.65). The segmentations generated by the ensembled model produced a mean Dice coefficient of 0.62 (IQR: 0.47 to 0.85) on the test set. The sensitivity for detecting a tumor, as defined by correctly predicting at least one voxel within a ground truth tumor, was 93.2%, and the Dice coefficient for the scans with correctly identified lesions was 0.67 (IQR: 0.53 to 0.85). In comparison, the mean interobserver Dice coefficient for the three physicians on the test set was 0.76 (IQR: 0.70 to 0.84). We observed strong correlation between physician-determined tumor size and model-predicted tumor size (Pearson correlation, r = 0.69, P < 0.0001).An end-to-end deep learning-based model was able to identify and segment lung tumors in a completely automated fashion, with near-expert level performance. Such models could soon be useful for clinical contouring and automatic quantification of tumor burden.