Beyond accuracy: Quantifying the reliability of multiple instance learning for whole slide image classification

Title (eng)
Beyond accuracy: Quantifying the reliability of multiple instance learning for whole slide image classification
Author
Hassan Keshvarikhojasteh
Author
Marc Aubreville
Author
Christof A. Bertram
Author
Josien P. W. Pluim
Author
Mitko Veta
Abstract (eng)
Machine learning models have become integral to many fields, but their reliability, defined as producing dependable, trustworthy, and domain-consistent predictions, remains a critical concern. Multiple Instance Learning (MIL) models designed for Whole Slide Image (WSI) classification in computational pathology are rarely evaluated in terms of reliability, leaving a key gap in understanding their suitability for high-stakes applications like clinical decision-making. In this paper, we address this gap by introducing three quantitative metrics for reliability assessment and applying them to several widely used MIL architectures across three region-wise annotated pathology datasets. Our findings indicate that the mean pooling instance (MEAN-POOL-INS) model demonstrates superior reliability compared to other networks, despite its simple architectural design and computational efficiency. These findings underscore the need of reliability evaluation alongside predictive performance in MIL models and establish MEAN-POOL-INS as a strong, trustworthy baseline for future research.
Keywords (eng)
Machine LearningCancers And NeoplasmsSequence AlignmentBreast CancerTissue DistributionGenome AnnotationReliabilityLearning Curves
Type (eng)
journal article
Language
[eng]
Persistent identifier
https://phaidra.vetmeduni.ac.at/o:5089
DOI
10.1371/journal.pone.0337261
Is in series
Title (eng)
PLOS One
Volume
20
Issue
12
ISSN
1932-6203
Issued
2025
Number of pages
15
Publication
Public Library of Science
Version type (eng)
version of record (published version)
Date issued
2025
Access rights (eng)
open access
License
CC BY 4.0 International
Rights statement (eng)
© 2025 Keshvarikhojasteh et al.