Hitchhiker's guide to cancer-associated lymphoid aggregates in histology images: manual and deep learning-based quantification approaches
This work addresses the need for objective biomarkers in cancer prognosis and prediction, but it is incremental as it builds on existing deep learning methods for a specific medical imaging task.
The paper tackles the problem of quantifying lymphoid aggregates in cancer histology images by providing manual identification guidelines and introducing HookNet-TLS, a deep learning algorithm for automated detection, which aims to reduce variability from manual analysis.
Quantification of lymphoid aggregates including tertiary lymphoid structures with germinal centers in histology images of cancer is a promising approach for developing prognostic and predictive tissue biomarkers. In this article, we provide recommendations for identifying lymphoid aggregates in tissue sections from routine pathology workflows such as hematoxylin and eosin staining. To overcome the intrinsic variability associated with manual image analysis (such as subjective decision making, attention span), we recently developed a deep learning-based algorithm called HookNet-TLS to detect lymphoid aggregates and germinal centers in various tissues. Here, we additionally provide a guideline for using manually annotated images for training and implementing HookNet-TLS for automated and objective quantification of lymphoid aggregates in various cancer types.