< Back to previous page

Project

Next-generation pathology by MILAN: multiplex immunohistochemistry and advanced trainable image analysis

Pathology (or the cellular analysis of resected tissues) remains an essential tool to come to the right diagnosis for many diseases. However, the amount of information retrieved from histological samples is still very limited and hasn’t changed in the last 50 years. We propose to make a quantum leap in histopathology by introducing an automated multiplex immunohistofluorescent method (MILAN) based on the high-throughput screening of over 80 biomarkers in just one FFPE tissue section at single cell resolution. This image-based technology will enable pathologists and researchers to investigate phenotypic changes in single cells of diseased tissues at unprecedented resolution, providing superior tools to identify markers that predict responsiveness to therapy. The MILAN procedure involves a hybrid approach integrating laboratory and computational steps, which will be offered as an integrated service. All instrumentation for automation of the laboratory procedures were recently installed (supported by investments of the department of pathology, KUL-LRD and LKI) and will enable us to generate massive amounts of data. For the subsequent computational analysis of the resulting images (involving high-throughput image QC and alignment, cell segmentation, automatic cell type recognition and spatial/neighborhood analysis), commercial platforms lack the required tools and flexibility for maximal data extraction and interpretation. Therefore, we have been developing our own artificial intelligence-based software tools tailored to process and analyze pathology-grade images allowing us to go far beyond mere intensity marker measurements. Proof-of-concept of this approach has been gathered (e.g. multiplex-based prediction of immunotherapy response in melanoma), but the pipelines and user-interface do not yet match the required throughput, automation, speed and quality that is needed to provide next-generation pathology services to (inter)national researchers and companies. This project will therefore focus on the automation and optimization of the dedicated software algorithms and user-interface that will allow stakeholders to interpret complex disease states and biomarkers at unprecedented resolution.
Date:1 Jan 2020 →  31 Dec 2021
Keywords:multiplex-IHC, artifical intelligence, image analysis
Disciplines:General diagnostics