Ecologically sustainable benchmarking of AI models for histopathology

Abstract Deep learning (DL) holds great promise to improve medical diagnostics, including pathology. Current DL research mainly focuses on performance. DL implementation potentially leads to environmental consequences but approaches for assessment of both performance and carbon footprint are missing...

Full description

Saved in:
Bibliographic Details
Main Authors: Yu-Chia Lan, Martin Strauch, Pourya Pilva, Nikolas E. J. Schmitz, Alireza Vafaei Sadr, Leon Niggemeier, Huong Quynh Nguyen, David L. Hölscher, Tri Q. Nguyen, Jesper Kers, Roman D. Bülow, Peter Boor
Format: Article
Language:English
Published: Nature Portfolio 2024-12-01
Series:npj Digital Medicine
Online Access:https://doi.org/10.1038/s41746-024-01397-x
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846100940242812928
author Yu-Chia Lan
Martin Strauch
Pourya Pilva
Nikolas E. J. Schmitz
Alireza Vafaei Sadr
Leon Niggemeier
Huong Quynh Nguyen
David L. Hölscher
Tri Q. Nguyen
Jesper Kers
Roman D. Bülow
Peter Boor
author_facet Yu-Chia Lan
Martin Strauch
Pourya Pilva
Nikolas E. J. Schmitz
Alireza Vafaei Sadr
Leon Niggemeier
Huong Quynh Nguyen
David L. Hölscher
Tri Q. Nguyen
Jesper Kers
Roman D. Bülow
Peter Boor
author_sort Yu-Chia Lan
collection DOAJ
description Abstract Deep learning (DL) holds great promise to improve medical diagnostics, including pathology. Current DL research mainly focuses on performance. DL implementation potentially leads to environmental consequences but approaches for assessment of both performance and carbon footprint are missing. Here, we explored an approach for developing DL for pathology, which considers both diagnostic performance and carbon footprint, calculated as CO2 or equivalent emissions (CO2eq). We evaluated various DL architectures used in computational pathology, including a large foundation model, across two diagnostic tasks of low and high complexity. We proposed a metric termed ‘environmentally sustainable performance’ (ESPer), which quantitatively integrates performance and operational CO2eq during training and inference. While some DL models showed comparable diagnostic performance, ESPer enabled prioritizing those with less carbon footprint. We also investigated how data reduction approaches can improve the ESPer of individual models. This study provides an approach facilitating the development of environmentally friendly, sustainable medical AI.
format Article
id doaj-art-f08a737b92404f7d9aca3eff1b3b96f7
institution Kabale University
issn 2398-6352
language English
publishDate 2024-12-01
publisher Nature Portfolio
record_format Article
series npj Digital Medicine
spelling doaj-art-f08a737b92404f7d9aca3eff1b3b96f72024-12-29T12:48:27ZengNature Portfolionpj Digital Medicine2398-63522024-12-017111110.1038/s41746-024-01397-xEcologically sustainable benchmarking of AI models for histopathologyYu-Chia Lan0Martin Strauch1Pourya Pilva2Nikolas E. J. Schmitz3Alireza Vafaei Sadr4Leon Niggemeier5Huong Quynh Nguyen6David L. Hölscher7Tri Q. Nguyen8Jesper Kers9Roman D. Bülow10Peter Boor11Institute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityDepartment of Pathology, University Medical Centre UtrechtDepartment of Pathology, Amsterdam UMC, University of AmsterdamInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityInstitute of Pathology, University Clinic Aachen, RWTH Aachen UniversityAbstract Deep learning (DL) holds great promise to improve medical diagnostics, including pathology. Current DL research mainly focuses on performance. DL implementation potentially leads to environmental consequences but approaches for assessment of both performance and carbon footprint are missing. Here, we explored an approach for developing DL for pathology, which considers both diagnostic performance and carbon footprint, calculated as CO2 or equivalent emissions (CO2eq). We evaluated various DL architectures used in computational pathology, including a large foundation model, across two diagnostic tasks of low and high complexity. We proposed a metric termed ‘environmentally sustainable performance’ (ESPer), which quantitatively integrates performance and operational CO2eq during training and inference. While some DL models showed comparable diagnostic performance, ESPer enabled prioritizing those with less carbon footprint. We also investigated how data reduction approaches can improve the ESPer of individual models. This study provides an approach facilitating the development of environmentally friendly, sustainable medical AI.https://doi.org/10.1038/s41746-024-01397-x
spellingShingle Yu-Chia Lan
Martin Strauch
Pourya Pilva
Nikolas E. J. Schmitz
Alireza Vafaei Sadr
Leon Niggemeier
Huong Quynh Nguyen
David L. Hölscher
Tri Q. Nguyen
Jesper Kers
Roman D. Bülow
Peter Boor
Ecologically sustainable benchmarking of AI models for histopathology
npj Digital Medicine
title Ecologically sustainable benchmarking of AI models for histopathology
title_full Ecologically sustainable benchmarking of AI models for histopathology
title_fullStr Ecologically sustainable benchmarking of AI models for histopathology
title_full_unstemmed Ecologically sustainable benchmarking of AI models for histopathology
title_short Ecologically sustainable benchmarking of AI models for histopathology
title_sort ecologically sustainable benchmarking of ai models for histopathology
url https://doi.org/10.1038/s41746-024-01397-x
work_keys_str_mv AT yuchialan ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT martinstrauch ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT pouryapilva ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT nikolasejschmitz ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT alirezavafaeisadr ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT leonniggemeier ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT huongquynhnguyen ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT davidlholscher ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT triqnguyen ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT jesperkers ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT romandbulow ecologicallysustainablebenchmarkingofaimodelsforhistopathology
AT peterboor ecologicallysustainablebenchmarkingofaimodelsforhistopathology