Page 118 - TD-4-3

Tumor Discovery

LETTER TO EDITOR
Redefining the role of radiation oncologists in

the AI era

Melek Yakar*
Department of Radiation Oncology, Faculty of Medicine, Osmangazi University, Eskişehir, Turkey

Dear Editor,
Recent years have seen significant acceleration in the integration of artificial intelligence
(AI) into radiation oncology practice. From automated contouring to treatment
planning optimization and big data analytics, AI offers remarkable advantages in terms
of efficiency and accuracy. However, this rapid transformation has also begun to redefine
the role of radiation oncologists in clinical decision-making. In this letter, I aim to
highlight the potential risk of physicians becoming distanced from critical decisions –
and even professionally isolated – as AI assumes a larger role, while also emphasizing the
importance of preserving core values such as ethics, autonomy, and empathy throughout
this transition.
While AI-driven systems have demonstrated superior performance in several
aspects of radiation oncology – such as patient assessment, clinical decision-making,
segmentation, dose prediction, and outcome modeling – it is essential to recognize that
these tools are only as reliable as the data they are trained on. Radiation oncologists
must continue to utilize their expertise in clinical guidelines, patient data, and
multidisciplinary assessments to make the best treatment decisions. Each individual’s
situation is unique, and the decision-making process requires both big data analysis and
*Corresponding author: clinical experience due to disease-specific factors. AI-powered systems can accelerate
Melek Yakar and support doctors’ decisions by providing recommendations based on clinical
(myakar@ogu.edu.tr) guidelines. The advantages of clinical decision support systems are that they reduce the
Citation: Yakar M. Redefining the margin of error by assisting doctors, analyzing complex patient data more effectively,
role of radiation oncologists in the and recommending the best treatment options for each patient, all while accelerating
AI era. Tumor Discov. 2025;4(3):110-
112. decision-making in multidisciplinary workflows. However, clinical decisions should
doi: 10.36922/TD025200039 not be left entirely to AI. Human oversight is essential to ensure the best outcomes and
further validation is needed for clinical acceptance. 1,2
Received: May 15, 2025
Accepted: May 22, 2025 One of the most time-consuming steps in treatment planning is organ and target
volume segmentation. AI-based methods, such as U-Net/TransU-Net convolutional
Published online: June 10, 2025 neural network models, have been utilized to reduce segmentation time, making it much
Copyright: © 2025 Author(s). faster than manual processes, which can take hours. Despite these advancements, full
3-6
This is an Open-Access article automation is still not possible. Clinical validation is required, and manual corrections
distributed under the terms of the
Creative Commons Attribution may still be necessary in certain cases, such as low-contrast tumors. In addition, clinical
License, permitting distribution, integration of these systems requires time, training, and standardization of evaluation
and reproduction in any medium, criteria. 7
provided the original work is
properly cited. Synthetic computed tomography (CT) images, generated from magnetic resonance
Publisher’s Note: AccScience imaging data using AI-based algorithms, are becoming increasingly utilized in
Publishing remains neutral with radiotherapy planning. Synthetic CT provides an alternative to conventional CT by
regard to jurisdictional claims in
published maps and institutional offering accurate electron density information, which reduces radiation exposure and
affiliations. improves workflow efficiency. However, challenges such as inter-vendor variability and

Volume 4 Issue 3 (2025) 110 doi: 10.36922/TD025200039