CAMEL: A Weakly Supervised Learning Framework for Histopathology Image Segmentation

Citations Over TimeTop 10% of 2019 papers

Abstract

Histopathology image analysis plays a critical role in cancer diagnosis and treatment. To automatically segment the cancerous regions, fully supervised segmentation algorithms require labor-intensive and time-consuming labeling at the pixel level. In this research, we propose CAMEL, a weakly supervised learning framework for histopathology image segmentation using only image-level labels. Using multiple instance learning (MIL)-based label enrichment, CAMEL splits the image into latticed instances and automatically generates instance-level labels. After label enrichment, the instance-level labels are further assigned to the corresponding pixels, producing the approximate pixel-level labels and making fully supervised training of segmentation models possible. CAMEL achieves comparable performance with the fully supervised approaches in both instance-level classification and pixel-level segmentation on CAMELYON16 and a colorectal adenoma dataset. Moreover, the generality of the automatic labeling methodology may benefit future weakly supervised learning studies for histopathology image analysis.

Related Papers

• → Bad pixel identification by means of principal components analysis(2002)45 cited
• → Low complexity photo sensor dead pixel detection algorithm(2012)14 cited
• → Sub-pixel mapping based on sub-pixel to sub-pixel spatial attraction model(2011)12 cited
• → Reversible Data Hiding Based on Multiple Strategies(2022)1 cited
• → The Big Pixel Microworld(2021)