Article Impact Level: HIGH Data Quality: STRONG Summary of The Lancet Oncology, 25(11), e581–e588. https://doi.org/10.1016/S1470-2045(24)00316-4 Dr. Javier E. Villanueva-Meyer et al.
Points
- AI advancements in neuro-oncology enhance brain cancer diagnosis, prognostication, and therapy, particularly through AI models targeting genomic markers, predicting treatment response, and distinguishing between disease progression and treatment effects.
- A recent Policy Review outlines AI’s potential to automate response assessments, improving accuracy and efficiency compared to traditional methods.
- New international guidelines recommend standardized AI methods to improve reliability in clinical trials and ensure patient safety, addressing the subjectivity in traditional tumor assessments.
- The guidelines call for AI models trained on large, diverse data sets, adherence to WHO criteria for tumor imaging, and precise image segmentation for analysis.
- Although AI shows promise, the review emphasizes the need for further research across diverse patient populations to optimize AI accuracy and efficacy in neuro-oncology.
Summary
The rapid advancement of artificial intelligence (AI) in neuro-oncology has led to significant progress in improving diagnosis, prognostication, and therapy for brain cancer patients. A recent Policy Review critically assesses the current applications of AI in neuro-oncology, highlighting the use of AI for diagnostic models targeting key genomic markers, predictive models for treatment response, and differentiation between true disease progression and treatment-related changes. This review also discusses the potential for AI to automate response assessments, an area that presents challenges in current clinical practice. By addressing these issues, AI is set to revolutionize how brain cancer is diagnosed and treated, offering more accurate and efficient methods than traditional approaches.
An international team of neuro-oncology experts has developed new guidelines for using AI in clinical practice to improve the reliability of clinical trial results and ensure patient safety. These guidelines call for standardized and validated AI methods in neuro-oncology, particularly to address the subjective nature of traditional tumor assessments conducted by radiologists. Lead author Spyridon Bakas emphasized the importance of AI in providing more objective measurements of tumor size, type, grade, and progression, offering a more consistent basis for treatment decisions. The recommendations aim to standardize AI usage across institutions, ensuring consistent and accurate diagnosis and treatment planning.
The guidelines stress the importance of using AI models developed with large and diverse patient data sets, ensuring tumor images are processed according to World Health Organization criteria, and properly segmenting images before analysis. These recommendations, presented at several major oncology meetings, highlight the necessity of ongoing research to refine AI tools for cancer care. Despite AI’s promising potential, the authors stress that more extensive studies on diverse patient populations are needed to further enhance the accuracy and effectiveness of these technologies in improving patient outcomes.
Link to the article: https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(24)00316-4/abstract
References Villanueva-Meyer, J. E., Bakas, S., Tiwari, P., Lupo, J. M., Calabrese, E., Davatzikos, C., Bi, W. L., Ismail, M., Akbari, H., Lohmann, P., Booth, T. C., Wiestler, B., Aerts, H. J. W. L., Rasool, G., Tonn, J. C., Nowosielski, M., Jain, R., Colen, R. R., Pati, S., … Galldiks, N. (2024). Artificial Intelligence for Response Assessment in Neuro Oncology (Ai-rano), part 1: Review of current advancements. The Lancet Oncology, 25(11), e581–e588. https://doi.org/10.1016/S1470-2045(24)00316-4
