Navigation

Discover expert commentary on essential clinical research in radiation oncology and artificial intelligence

Subscribe

Managing Editors

  • slide 2
    Marc Kohli, MD

    Medical Director of Imaging Informatics

    Associate Professor of Radiology and Biomedical Imaging UCSF

Created by physicians and scientists, ACR Journal Advisor provides access to compelling radiation oncology and artificial intelligence research and expert interpretation. First established in 1995 as a simple, email-based service to optimize clinical decision-making, it has since grown into an international online medical library of seminal literature, helpful reference tools, and incisive interpretations from trusted and renowned radiation oncologists.

These services are made possible by the generous contributions of the National Library of Medicine, our editorial board, the American College of Radiology (ACR), and our sponsors. Guided by its mission of service to the radiation oncology community, the ACR is working to increase the value of this resource with robust educational, technical, and organizational support.

If you’re not already a subscriber, take the next step and register to begin filling your research library with critical new publications and expert editorial commentary.

Subscribers recieve monthly Editor's Choice commentaries

Radiation Oncology

Editor or Editors: Gregory Videtic, M.D.

Topic | SubTopic: Lung and Mediastinum | NSCLCa

Article: Palma DA,et al.,Measuring the integration of Stereotactic Ablative Radiothera..

Comment By: Gregory Videtic, M.D.,


This study that is quite remarkable for the fact that it was feasible to carry it out and so its authors deservedly should be commended for this publication. Simply put, this was a phase II study in early-stage operable lung cancer patients who underwent preoperative stereotactic body radiotherapy (SBRT, or stereotactic ablative radiotherapy, SABR) to determine the pathologic rate of complete responses (pCR) to the SBRT and to assess oncologic and toxicity outcomes after a combined approach of neoadjuvant radiotherapy followed by surgery, which was carried out 10 weeks after the completion of SBRT. Their results showed a 60% pCR rate which was lower than hypothesized and no untoward complications from the additive therapies. In their discussion, the authors try to reconcile the available clinical data which suggest high local control rates after SBRT and the pCR rates seen. They note that one explanation might be that resection at 10 weeks followed by pathological assessment is too early for the full expression of SBRT-related killing to be measured. The second explanation is that clinicians are overestimating the effects of SBRT on local control since measuring it after SBRT is fraught with the challenges of interpreting post-SBRT lung changes. In light of these results, they caution about the use of SBRT inoperable patients.

With respect to their findings, I support their suggestion of a “too early” assessment of the pCR rated likely as the most appropriate explanation for their lower than expected response rate. The accumulating SBRT data from both retrospective and prospective series is compelling in that regard, including published long term follow up (e.g., 5 years and greater) clinical data from prospective studies, the demonstration of prolonged effect of SBRT as seen in studies on late PET avidity at treated sites, and provocative SBRT overall survival outcomes comparable to surgical results for equivalent clinically staged patients.

Artificial Intelligence

Editor or Editors: Judy Gichoya, MBchB; Marc Kohli, MD

Topic: Ethics of Artificial Intelligence

Article: Ethics of Artificial Intelligence in Radiology: Summary of the Joint European and North American Multisociety Statement

Comment By: Judy Gichoya, MBchB; Marc Kohli, MD


Several articles have been written on Ethics of AI, but this article is an important addition to the artificial intelligence (AI) literature because it was designed to initiate a dialogue among radiologists for creating an ethical framework for implementation of AI in radiology. This paper obtained inputs from different radiology societies in North America and Europe; with involvement of radiologists, legal experts and patient advocates. While the paper does not presume to provide answers to the many moral and ethical issues surrounding the use of AI, it does raise key, fundamental questions that will need to be addressed by the radiology community to steer technological development in a direction that is optimal clinically while maintaining the rights of the patient. The paper is mirrored across three main pillars raising questions on ethics of data, algorithm and implementation. For example: Are we able to explain how our AI is making decisions and is bias being introduced in the decision-making? How will we minimize the risk of patient harm from malicious attacks and privacy breaches? How will we evaluate trained models for clinical effectiveness, ethical behavior, and security? How will we monitor AI models in clinical workflow to ensure they perform as predicted and that performance does not degrade over time? Financial, legal, and regulatory issues are also raised. The article does an excellent job of outlining a framework that the radiology community needs to use to develop more prescriptive code of ethics and practice for AI. By recognizing that success of AI in radiology will involve more than radiologists, this paper had a public input period that serves as a good point for tackling ethical issues around development and use of AI in radiology.

ad