An interpretable automated detection system for FISH-based HER2 - - PowerPoint PPT Presentation

An interpretable automated detection system for FISH-based HER2 oncogene amplification testing in histo-pathological routine images of breast and gastric cancer diagnostics

Sarah Schmell, Falk Zakrzewski, Walter de Back, Martin Weigert, Uwe Schmidt, Torsten Wenke, Silke Zeugner, Robert Mantey, Christian Sperling, Ingo Roeder, Pia Hoenscheid, Daniela Aust and Gustavo Baretton MIDL 2020 – Short Paper

Pathologist

Complex Diagnostics and Increasing Workload

1

+

AI assisted Pathologist

2 2 2 3 5 4 4 2 3 3 2 2 3 3 3 1 2 2 3 2 2 2 3 4 2 2

# HER2 # CEP17 per nucleus

FISH HER2 amplification testing

< 2.0 → HER2 negative → Not HER2-targeted therapy ≥ 2.0 → HER2 positive low → HER2-targeted therapy > 5.0 → HER2 positive high

• 1. Step: Nucleus Filtering with the Nucleus Detector

2 Schmidt et al. (2018)

HER2 CEP17

Input (1600 × 1200 px) Segmentation of detected Nuclei Output (~80 × 80 px)

• 2. Step: Grading Nuclei with the Nucleus Classifier

3

Artifact Background HER2 normal expression Artifact HER2 low amplification HER2 high amplification …

Output (Grading and CAM) Input (~80 × 80 px) Postprocessing (image-wide Grading) HER2 low amplification

• 3. Step: Second Opinion from the Signal Detector

4 Lin et al. (2017)

Artifact HER2 normal expression HER2 low amplification Artifact HER2 low amplification HER2 high amplification …

Output (Grading and BBoxes) Input (~80 × 80 px) Postprocessing (image-wide Grading) HER2 low amplification

• 4. Step: Reviewing the Report

5

Enhancements to the AI assistance

THANK YOU

An interpretable automated detection system for FISH-based HER2 oncogene amplification testing in histo-pathological routine images of breast and gastric cancer diagnostics