Beyond Occlusion: In Search for Near Real-Time Explainability
of CNN-Based Prostate Cancer Classification

D 2025

Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification

KREBS, Martin; Jan OBDRŽÁLEK; Vít MUSIL a Tomáš BRÁZDIL

Základní údaje

Originální název

Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification

Autoři

KREBS, Martin; Jan OBDRŽÁLEK; Vít MUSIL a Tomáš BRÁZDIL

Vydání

NEW YORK, 2025 IEEE 22nd International Symposium on Biomedical Imaging, ISBI, od s. 1-5, 5 s. 2025

Nakladatel

IEEE

Další údaje

Jazyk

angličtina

Typ výsledku

Stať ve sborníku

Obor

10200 1.2 Computer and information sciences

Utajení

není předmětem státního či obchodního tajemství

Forma vydání

elektronická verze "online"

Odkazy

Published version, Preprint

Organizační jednotka

Fakulta informatiky

ISBN

979-8-3315-2053-3

ISSN

DOI

https://doi.org/10.1109/ISBI60581.2025.10980802

UT WoS

001546451000151

EID Scopus

2-s2.0-105005832793

Klíčová slova anglicky

digital histopathology; prostate cancer; convolutional neural networks; explainable AI; artificial intelligence

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 5. 1. 2026 13:49, RNDr. Vít Musil, Ph.D.

Anotace

V originále

Deep neural networks are starting to show their worth in critical applications such as assisted cancer diagnosis. However, for their outputs to get accepted in practice, the results they provide should be explainable in a way easily understood by pathologists. A well-known and widely used explanation technique is occlusion, which, however, can take a long time to compute, thus slowing the development and interaction with pathologists. In this work, we set out to find a faster replacement for occlusion in a successful system for detecting prostate cancer. Since there is no established framework for comparing the performance of various explanation methods, we first identified suitable comparison criteria and selected corresponding metrics. Based on the results, we were able to choose a different explanation method, which cut the previously required explanation time at least by a factor of 10, without any negative impact on the quality of outputs. This speedup enables rapid iteration in model development and debugging and brings us closer to adopting AI-assisted prostate cancer detection in clinical settings. We propose that our approach to finding the replacement for occlusion can be used to evaluate candidate methods in other related applications.

Návaznosti

101079183, interní kód MU

Název: BioMedAI TWINNING

Investor: Evropská unie, BioMedAI TWINNING, Rozšiřování účasti a posílení ERA

90140, velká výzkumná infrastruktura

Název: e-INFRA CZ

Přiložené soubory

Beyond_Occlusion_In_Search_for_Near_Real-Time_Explainability_of_CNN-Based_Prostate_Cancer_Classification.pdf

Požádat o autorskou verzi souboru

Citovat

KREBS, Martin; Jan OBDRŽÁLEK; Vít MUSIL a Tomáš BRÁZDIL. Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification. Online. In 2025 IEEE 22nd International Symposium on Biomedical Imaging, ISBI. NEW YORK: IEEE, 2025, s. 1-5. ISBN 979-8-3315-2053-3. Dostupné z: https://doi.org/10.1109/ISBI60581.2025.10980802.

@inproceedings{2544119,
   author = {Krebs, Martin and Obdržálek, Jan and Musil, Vít and Brázdil, Tomáš},
   address = {NEW YORK},
   booktitle = {2025 IEEE 22nd International Symposium on Biomedical Imaging, ISBI},
   doi = {https://doi.org/10.1109/ISBI60581.2025.10980802},
   keywords = {digital histopathology; prostate cancer; convolutional neural networks; explainable AI; artificial intelligence},
   howpublished = {elektronická verze "online"},
   language = {eng},
   location = {NEW YORK},
   isbn = {979-8-3315-2053-3},
   pages = {1-5},
   publisher = {IEEE},
   title = {Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification},
   url = {https://ieeexplore.ieee.org/document/10980802},
   year = {2025}
}

TY  - CONF
ID  - 2544119
AU  - Krebs, Martin - Obdržálek, Jan - Musil, Vít - Brázdil, Tomáš
PY  - 2025
TI  - Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification
PB  - IEEE
CY  - NEW YORK
SN  - 9798331520533
KW  - digital histopathology
KW  - prostate cancer
KW  - convolutional neural networks
KW  - explainable AI
KW  - artificial intelligence
UR  - https://ieeexplore.ieee.org/document/10980802
N2  - Deep neural networks are starting to show their worth in critical applications such as assisted cancer diagnosis. However, for their outputs to get accepted in practice, the results they provide should be explainable in a way easily understood by pathologists. A well-known and widely used explanation technique is occlusion, which, however, can take a long time to compute, thus slowing the development and interaction with pathologists. In this work, we set out to find a faster replacement for occlusion in a successful system for detecting prostate cancer. Since there is no established framework for comparing the performance of various explanation methods, we first identified suitable comparison criteria and selected corresponding metrics. Based on the results, we were able to choose a different explanation method, which cut the previously required explanation time at least by a factor of 10, without any negative impact on the quality of outputs. This speedup enables rapid iteration in model development and debugging and brings us closer to adopting AI-assisted prostate cancer detection in clinical settings. We propose that our approach to finding the replacement for occlusion can be used to evaluate candidate methods in other related applications.
ER  -

KREBS, Martin; Jan OBDRŽÁLEK; Vít MUSIL a Tomáš BRÁZDIL. Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification. Online. In \textit{2025 IEEE 22nd International Symposium on Biomedical Imaging, ISBI}. NEW YORK: IEEE, 2025, s.~1-5. ISBN~979-8-3315-2053-3. Dostupné z: https://doi.org/10.1109/ISBI60581.2025.10980802.

Přehled o publikaci