Nature Communications

Pathologist-like explainable AI for interpretable Gleason grading in prostate cancer

G Gesa Mittmann S Sara Laiouar-Pedari H Hendrik A. Mehrtens S Sarah Haggenmüller T Tabea-Clara Bucher T Tirtha Chanda N Nadine T. Gaisa M Mathias Wagner G Gilbert Georg Klamminger T Tilman T. Rau C Christina Neppl E Eva Maria Compérat A Andreas Gocht M Monika Haemmerle N Niels J. Rupp J Jula Westhoff I Irene Krücken M Maximilian Seidl C Christian M. Schürch M Marcus Bauer W Wiebke Solass Y Yu Chun Tam F Florian Weber R Rainer Grobholz J Jaroslaw Augustyniak T Thomas Kalinski C Christian Hörner K Kirsten D. Mertz C Constanze Döring A Andreas Erbersdobler G Gabriele Deubler F Felix Bremmer U Ulrich Sommer M Michael Brodhun J Jon Griffin M Maria Sarah L. Lenon K Kiril Trpkov L Liang Cheng (Institute of Functional Nano & Soft Materials (FUNSOM), Jiangsu Key Laboratory for Carbon-Based Functional Materials and Devices) F Fei Chen A Angelique Levi G Guoping Cai T Tri Q. Nguyen A Ali Amin A Alessia Cimadamore A Ahmed Shabaik V Varsha Manucha N Nazeel Ahmad N Nidia Messias F Francesca Sanguedolce D Diana Taheri E Ezra Baraban L Liwei Jia R Rajal B. Shah F Farshid Siadat N Nicole Swarbrick K Kyung Park O Oudai Hassan S Siamak Sakhaie M Michelle R. Downes H Hiroshi Miyamoto S Sean R. Williamson T Tim Holland-Letz C Christoph Wies C Carolin V. Schneider J Jakob Nikolas Kather Y Yuri Tolkach T Titus J. Brinker

Abstract

Abstract The aggressiveness of prostate cancer is primarily assessed from histopathological data using the Gleason scoring system. Conventional artificial intelligence (AI) approaches can predict Gleason scores, but often lack explainability, which may limit clinical acceptance. Here, we present an alternative, inherently explainable AI that circumvents the need for post-hoc explainability methods. The model was trained on 1,015 tissue microarray core images, annotated with detailed pattern descriptions by 54 international pathologists following standardized guidelines. It uses pathologist-defined terminology and was trained using soft labels to capture data uncertainty. This approach enables robust Gleason pattern segmentation despite high interobserver variability. The model achieved comparable or superior performance to direct Gleason pattern segmentation (Dice score: $${0.713}_{\pm 0.003}$$ 0.713 ± 0.003 vs. $${0.691}_{\pm 0.010}$$ 0.691 ± 0.010 ) while providing interpretable outputs. We release this dataset to encourage further research on segmentation in medical tasks with high subjectivity and to deepen insights into pathologists’ reasoning.

Download PDF View Original
APA MLA Chicago BibTeX RIS

Article Details

Volume / Issue Vol. 16, Issue 1
Published October 08, 2025
ISSN 2041-1723
Publisher Nature Portfolio

Journal Info

Nature Communications

Nature Portfolio
ISSN: 2041-1723 Open Access Life Sciences

Authors (67)

G

Gesa Mittmann

S

Sara Laiouar-Pedari

H

Hendrik A. Mehrtens

S

Sarah Haggenmüller

T

Tabea-Clara Bucher

T

Tirtha Chanda

N

Nadine T. Gaisa

M

Mathias Wagner

G

Gilbert Georg Klamminger

T

Tilman T. Rau

C

Christina Neppl

E

Eva Maria Compérat

A

Andreas Gocht

M

Monika Haemmerle

N

Niels J. Rupp

J

Jula Westhoff

I

Irene Krücken

M

Maximilian Seidl

C

Christian M. Schürch

M

Marcus Bauer

W

Wiebke Solass

Y

Yu Chun Tam

F

Florian Weber

R

Rainer Grobholz

J

Jaroslaw Augustyniak

T

Thomas Kalinski

C

Christian Hörner

K

Kirsten D. Mertz

C

Constanze Döring

A

Andreas Erbersdobler

G

Gabriele Deubler

F

Felix Bremmer

U

Ulrich Sommer

M

Michael Brodhun

J

Jon Griffin

M

Maria Sarah L. Lenon

K

Kiril Trpkov

L

Liang Cheng

Institute of Functional Nano & Soft Materials (FUNSOM), Jiangsu Key Laboratory for Carbon-Based Functional Materials and Devices

F

Fei Chen

A

Angelique Levi

G

Guoping Cai

T

Tri Q. Nguyen

A

Ali Amin

A

Alessia Cimadamore

A

Ahmed Shabaik

V

Varsha Manucha

N

Nazeel Ahmad

N

Nidia Messias

F

Francesca Sanguedolce

D

Diana Taheri

E

Ezra Baraban

L

Liwei Jia

R

Rajal B. Shah

F

Farshid Siadat

N

Nicole Swarbrick

K

Kyung Park

O

Oudai Hassan

S

Siamak Sakhaie

M

Michelle R. Downes

H

Hiroshi Miyamoto

S

Sean R. Williamson

T

Tim Holland-Letz

C

Christoph Wies

C

Carolin V. Schneider

J

Jakob Nikolas Kather

Y

Yuri Tolkach

T

Titus J. Brinker

Related Articles from this Journal

Synthesis of adaptive 15N-nitrosyl-Co7 nanocluster for electrocatalytic C–H functionalization

Chao Wu, Xu Zhang et al. Aug 2026 10.1038/s41467-026-76516-1

Pan-Ebolavirus nanoparticle vaccine provides protection in rodents from lethal infection by Zaire and Sudan viruses

Connor Weidle, Natalie Brunette et al. Aug 2026 10.1038/s41467-026-76114-1

Multi-population GWAS meta-analysis identifies bladder cancer susceptibility loci and highlights genetic regulation of smoking-related risk

Ludmila Prokunina-Olsson, Oscar Florez-Vargas et al. Aug 2026 10.1038/s41467-026-76157-4

Genome-wide annotation of human multi-nucleotide variants reveals widespread functional differences from single nucleotide variants

Weiwei Jin, Wen Cao et al. Aug 2026 10.1038/s41467-026-76470-y

Allele-specific chromatin architecture shapes imprinted domains and coordinates a distal enhancer and antisense transcription at the mouse Mest-Copg2 domain

Bongmin Bae, Katherine Gu et al. Aug 2026 10.1038/s41467-026-76506-3