Part-Prototype Models in Medical Imaging: Applications and Current Challenges

Recent developments in Artificial Intelligence have increasingly focused on explainability research. The potential of Explainable Artificial Intelligence (XAI) in producing trustworthy computer-aided diagnosis systems and its usage for knowledge discovery are gaining interest in the medical imaging...

Full description

Saved in:
Bibliographic Details
Main Authors: Lisa Anita De Santi, Franco Italo Piparo, Filippo Bargagna, Maria Filomena Santarelli, Simona Celi, Vincenzo Positano
Format: Article
Language:English
Published: MDPI AG 2024-10-01
Series:BioMedInformatics
Subjects:
Online Access:https://www.mdpi.com/2673-7426/4/4/115
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846105651787333632
author Lisa Anita De Santi
Franco Italo Piparo
Filippo Bargagna
Maria Filomena Santarelli
Simona Celi
Vincenzo Positano
author_facet Lisa Anita De Santi
Franco Italo Piparo
Filippo Bargagna
Maria Filomena Santarelli
Simona Celi
Vincenzo Positano
author_sort Lisa Anita De Santi
collection DOAJ
description Recent developments in Artificial Intelligence have increasingly focused on explainability research. The potential of Explainable Artificial Intelligence (XAI) in producing trustworthy computer-aided diagnosis systems and its usage for knowledge discovery are gaining interest in the medical imaging (MI) community to support the diagnostic process and the discovery of image biomarkers. Most of the existing XAI applications in MI are focused on interpreting the predictions made using deep neural networks, typically including attribution techniques with saliency map approaches and other feature visualization methods. However, these are often criticized for providing incorrect and incomplete representations of the black-box models’ behaviour. This highlights the importance of proposing models intentionally designed to be self-explanatory. In particular, part-prototype (PP) models are interpretable-by-design computer vision (CV) models that base their decision process on learning and identifying representative prototypical parts from input images, and they are gaining increasing interest and results in MI applications. However, the medical field has unique characteristics that could benefit from more advanced implementations of these types of architectures. This narrative review summarizes existing PP networks, their application in MI analysis, and current challenges.
format Article
id doaj-art-a592cd0bb066421b90cb7a66f2d4618c
institution Kabale University
issn 2673-7426
language English
publishDate 2024-10-01
publisher MDPI AG
record_format Article
series BioMedInformatics
spelling doaj-art-a592cd0bb066421b90cb7a66f2d4618c2024-12-27T14:13:18ZengMDPI AGBioMedInformatics2673-74262024-10-01442149217210.3390/biomedinformatics4040115Part-Prototype Models in Medical Imaging: Applications and Current ChallengesLisa Anita De Santi0Franco Italo Piparo1Filippo Bargagna2Maria Filomena Santarelli3Simona Celi4Vincenzo Positano5Department of Information Engineering, University of Pisa, 56122 Pisa, ItalyDepartment of Information Engineering, University of Pisa, 56122 Pisa, ItalyDepartment of Information Engineering, University of Pisa, 56122 Pisa, ItalyCNR Institute of Clinical Physiology, 56124 Pisa, ItalyBioengineering Unit, Fondazione Toscana G. Monasterio, 56124 Pisa, ItalyBioengineering Unit, Fondazione Toscana G. Monasterio, 56124 Pisa, ItalyRecent developments in Artificial Intelligence have increasingly focused on explainability research. The potential of Explainable Artificial Intelligence (XAI) in producing trustworthy computer-aided diagnosis systems and its usage for knowledge discovery are gaining interest in the medical imaging (MI) community to support the diagnostic process and the discovery of image biomarkers. Most of the existing XAI applications in MI are focused on interpreting the predictions made using deep neural networks, typically including attribution techniques with saliency map approaches and other feature visualization methods. However, these are often criticized for providing incorrect and incomplete representations of the black-box models’ behaviour. This highlights the importance of proposing models intentionally designed to be self-explanatory. In particular, part-prototype (PP) models are interpretable-by-design computer vision (CV) models that base their decision process on learning and identifying representative prototypical parts from input images, and they are gaining increasing interest and results in MI applications. However, the medical field has unique characteristics that could benefit from more advanced implementations of these types of architectures. This narrative review summarizes existing PP networks, their application in MI analysis, and current challenges.https://www.mdpi.com/2673-7426/4/4/115deep learningXAIinterpretability-by-designpart-prototype modelsmedical imaging
spellingShingle Lisa Anita De Santi
Franco Italo Piparo
Filippo Bargagna
Maria Filomena Santarelli
Simona Celi
Vincenzo Positano
Part-Prototype Models in Medical Imaging: Applications and Current Challenges
BioMedInformatics
deep learning
XAI
interpretability-by-design
part-prototype models
medical imaging
title Part-Prototype Models in Medical Imaging: Applications and Current Challenges
title_full Part-Prototype Models in Medical Imaging: Applications and Current Challenges
title_fullStr Part-Prototype Models in Medical Imaging: Applications and Current Challenges
title_full_unstemmed Part-Prototype Models in Medical Imaging: Applications and Current Challenges
title_short Part-Prototype Models in Medical Imaging: Applications and Current Challenges
title_sort part prototype models in medical imaging applications and current challenges
topic deep learning
XAI
interpretability-by-design
part-prototype models
medical imaging
url https://www.mdpi.com/2673-7426/4/4/115
work_keys_str_mv AT lisaanitadesanti partprototypemodelsinmedicalimagingapplicationsandcurrentchallenges
AT francoitalopiparo partprototypemodelsinmedicalimagingapplicationsandcurrentchallenges
AT filippobargagna partprototypemodelsinmedicalimagingapplicationsandcurrentchallenges
AT mariafilomenasantarelli partprototypemodelsinmedicalimagingapplicationsandcurrentchallenges
AT simonaceli partprototypemodelsinmedicalimagingapplicationsandcurrentchallenges
AT vincenzopositano partprototypemodelsinmedicalimagingapplicationsandcurrentchallenges