Pathologist computer-aided diagnostic scoring of tumor cell fraction: A Swiss national study.

Ana Leni Frei, Raphaël Oberson, Elias Baumann, Aurel Perren, Rainer Grobholz, Alessandro Lugli, Heather Dawson, Christian Abbet, Ibai Lertxundi, Stefan Reinhard, Aart Mookhoek, Johann Feichtinger, Rossella Sarro, Gallus Gadient, Corina Dommann-Scherrer, Jessica Barizzi, Sabina Berezowska, Katharina Glatz, Susanne Dertinger, Yara Banz, Rene Schoenegg, Laura Rubbia-Brandt, Achim Fleischmann, Guenter Saile, Pierre Mainil-Varlet, Ruggero Biral, Luca Giudici, Alex Soltermann, Audrey Baur Chaubert, Sylvia Stadlmann, Joachim Diebold, Kristof Egervari, Charles Bénière, Francesca Saro, Andrew Janowczyk, Inti Zlobec

Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc 2023 Sep 22

filter terms:

Tumor cell fraction (TCF) estimation is a common clinical task with well-established large inter-observer variability. It thus provides an ideal testbed to evaluate potential impacts of employing a computer-aided diagnostic (TCFCAD) tool to support pathologists' evaluation. During a National Slide Seminar event, pathologists (n=69) were asked to visually estimate TCF in 10 regions of interest (ROI) from hematoxylin and eosin (H&E) colorectal cancer images intentionally curated for diverse tissue compositions, cellularity, and stain intensities. Next, they re-evaluated the same ROIs while being provided a TCFCAD created overlay highlighting predicted tumor versus non-tumor cells, together with the corresponding TCF percentage. Participants also reported confidence levels in their assessments using a 5-tiers scale, indicating no confidence to high confidence, respectively. The TCF ground truth (GT) was defined by manual cell-counting by experts. When assisted, inter-observer variability significantly decreased, showing estimates converging to the GT. This improvement remained even when TCFCAD predictions deviated slightly from the GT. The standard-deviation of estimated TCF to the GT across ROIs was 9.9% vs 5.8% with TCFCAD, p < 0.0001. The intraclass correlation coefficient increased from 0.8 to 0.93 (CI95% [0.65, 0.93] vs [0.86, 0.98]) and pathologists stated feeling more confident when aided (3.67 ± 0.81 vs. 4.17 ± 0.82 with CAD). TCFCAD estimation support demonstrated improved scoring accuracy, inter-pathologist agreement and scoring confidence. Interestingly, pathologists also expressed more willingness to use such a CAD tool at the end of the survey, highlighting the importance of training/education to increase adoption of CAD systems.Copyright © 2023. Published by Elsevier Inc.

Citation

PMID: 37742926

View Full Text