AI Outperforms People in Scoring Melanoma Tumor-Infiltrating Immune Cells

New research from Yale confirms that artificial intelligence (AI)-based scoring of melanoma tumor-infiltrating immune cells called lymphocytes significantly outperforms traditional pathologist eyeballing. The study, published in JAMA Network Open, found open-source AI tools offered a more standardized and reproducible method for assessment, underscoring the potential for AI to enhance clinical pathology workflows.

Tumor-infiltrating lymphocytes are used as a biomarker for melanoma, serving as indicators of how well the immune system is responding to the cancer. Having more tumor-infiltrating lymphocytes is associated with better outcomes for patients and tracking these immune cells can inform diagnoses and treatment decisions.

"Our findings suggest that an AI-driven lymphocyte quantification tool may provide consistent, reliable assessments with a strong potential for clinical use, offering a robust alternative to traditional methods,” says lead author Thazin Nwe Aung, PhD, associate research scientist in pathology at Yale School of Medicine (YSM).

The study was led by researchers at YSM and the Karolinska Institute in Sweden and included 45 institutions around the world.

AI excels at measuring melanoma-infiltrating lymphocytes

In the study, a total of 98 participants quantified tumor-infiltrating lymphocytes on 60 samples of melanoma tissue. Forty participants were pathologists who used traditional, visual methods to assess the tissue, while 11 pathologists and 47 non-pathologist scientists used AI.

The AI algorithm demonstrated superior reproducibility, significantly outperforming visual assessments.

Aung notes that even though the retrospective nature of the study limits demonstration of clinical use, “the publicly available dataset and open-source AI tool offer a foundation for future validation and integration into melanoma management.”

Aung, who works in the lab of David Rimm, MD, PhD, Anthony N. Brady Professor of Pathology at YSM, emphasizes the collaborative nature of the study: “I’m especially proud that 15 Yale School of Medicine faculty and staff contributed to this work. It’s a great example of how the Department of Pathology at Yale is leading the way in AI-driven pathology research.”

Other Yale researchers who worked on this study include Matthew Liu, BSc; David Su, MD; Matthew D. Vesely, MD, PhD; Yalai Bai, MD, PhD; Dijana Djureinovic, PhD, MSc; Pok Fai Wong, MD, PhD; Katherine Bates, BA; Nay Nwe Chan, PhD; Niki Gavrielatou, MD, PhD; Mengni He, MSc; Sneha Burela, MD; Shawn Cowper, MD; David Rimm, MD, PhD; and Goran Micevic, MD, PhD.

The research reported in this news article was supported by the National Institutes of Health (awards P50CA225450, U54CA263001, P50CA121974 to Yale SPORE in Skin Cancer, P50CA196530 to Yale SPORE in Lung Cancer, and P30CA016359 to Yale Cancer Center) and Yale University. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Additional support was provide by the Robert E. Leet and Clara Guthrie Patterson Trust, the Tower Cancer Research Foundation, and the Lion Heart Research Foundation, provided through the Yale School of Medicine.