Study
NLST
(Learn more about this study)
Project ID
NLST-177
Initial CDAS Request Approval
Nov 18, 2015
Title
Computer-Aided Diagnosis System for Histopathology Image Analysis
Summary
Histopathological examination of biopsy tissues is a powerful method for the prognosis of critical diseases such as breast, prostate, lung or ovarian cancers. Despite significant advances in high-speed and high-resolution scanning devices or in virtual exploration capabilities, the clinical analysis of whole slide images largely remains the work of human experts. At the hospital level, the task consists in the daily examination of images that directly impacts clinical diagnosis and treatment decisions. It is important for clinicians to have a more consistent and less subjective support tool for these tasks. The field of histopathology image analysis has a wide range of applications such as staining correction, large-scale image analysis, image segmentation, classification, retrieval etc.
Aims
We plan to develop an end-to-end learning framework that is able to interpret the specific multi-magnification composition of histopathology images in order to identify and segment semantically meaningful biomarkers for different cancer types.
The model should be able to identify, detect and segment regions that correspond to cancer from large tumor sections, and provide meaningful clinical descriptors of these cancerous regions (as a junior pathologist would do).
Our system was tested on ovarian carcinoma subtypes and we would like to validate it on larger cohorts and for different cancer types such as lung cancer.
Another part of our research would focus on the multi-modal analysis of lung tumours and the automatic detection of cancer from CT and Histopathology slides.
Collaborators
Medical Image Analysis Lab, Simon Fraser University.
