Explainable AI: A review of applications to neuroimaging data
Объяснимый искусственный интеллект: обзор приложений к данным нейровизуализации
2022-12-01
SCID: 54.1/5fue7775
Discuss with AI
Deep neural networksExplainable artificial intelligenceModel reliabilityNeuroimaging dataPost-hoc relevance techniques
Figures from the paper
Abstract (AI)
Deep neural networks (DNNs) have transformed the field of computer vision and currently constitute some of the best models for representations learned via hierarchical processing in the human brain. In medical imaging, these models have shown human-level performance and even higher in the early diagnosis of a wide range of diseases. However, the goal is often not only to accurately predict group membership or diagnose but also to provide explanations that support the model decision in a context that a human can readily interpret. The limited transparency has hindered the adoption of DNN algorithms across many domains. Numerous explainable artificial intelligence (XAI) techniques have been developed to peer inside the “black box” and make sense of DNN models, taking somewhat divergent approaches. Here, we suggest that these methods may be considered in light of the interpretation goal, including functional or mechanistic interpretations, developing archetypal class instances, or assessing the relevance of certain features or mappings on a trained model in a post-hoc capacity. We then focus on reviewing recent applications of post-hoc relevance techniques as applied to neuroimaging data. Moreover, this article suggests a method for comparing the reliability of XAI methods, especially in deep neural networks, along with their advantages and pitfalls.
Key Findings
1
Deep neural networks achieve human-level or higher performance in medical imaging for early diagnosis across diverse diseases, but limited transparency restricts adoption.
2
Explainable AI methods for deep networks pursue different interpretation goals, including functional or mechanistic understanding, archetypal class-instance generation, and post-hoc feature relevance assessment.
3
The article proposes a method for comparing the reliability of explainable AI methods, particularly for deep neural networks, while outlining their advantages and pitfalls.
4
The review focuses on recent applications of post-hoc relevance techniques to neuroimaging data.
Research Object
post-hoc explainable artificial intelligence techniques applied to neuroimaging data and deep neural network models
Research Subject
interpretability and reliability of deep neural network decisions, including feature or mapping relevance, functional or mechanistic explanations, and archetypal class instances
Publication Details
Publication Date
2022-12-01
Journal
Publisher
ISSN
Cited by
85
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai20
Very Deep Convolutional Networks for Large-Scale Image Recognition2014
A Unified Approach to Interpreting Model Predictions2017
Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement2009
The Cochrane Collaboration's tool for assessing risk of bias in randomised trials2011
The “Golden Age” of Probiotics: A Systematic Review and Meta-Analysis of Randomized and Observational Studies in Preterm Infants2017
Representation Learning: A Review and New Perspectives2013
Learning Deep Features for Discriminative Localization2016
Towards A Rigorous Science of Interpretable Machine Learning2017
Wide & Deep Learning for Recommender Systems2016
Deep learning with convolutional neural networks for EEG decoding and visualization2017
Scalable and accurate deep learning with electronic health records2018
Deep learning for healthcare: review, opportunities and challenges2017
A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI2020
Causability and explainability of artificial intelligence in medicine2019
An overview of deep learning in medical imaging focusing on MRI2018
Interactive machine learning for health informatics: when do we need the human-in-the-loop?2016
A Survey of Methods for Explaining Black Box Models2018
A multilayer multimodal detection and prediction model based on explainable artificial intelligence for Alzheimer’s disease2021
Layer-Wise Relevance Propagation for Explaining Deep Neural Network Decisions in MRI-Based Alzheimer's Disease Classification2019
"What is relevant in a text document?": An interpretable machine learning approach2017