Interpreting and Correcting Medical Image Classification with PIP-Net (2023-07-19T00:00:00.000000Z)

TL;DR

This paper is the first to show that humans can manually correct the reasoning of PIP-Net by directly disabling undesired prototypes, and conclude that part-prototype models are promising for medical applications due to their interpretability and potential for advanced model debugging.

Abstract

Part-prototype models are explainable-by-design image classifiers, and a promising alternative to black box AI. This paper explores the applicability and potential of interpretable machine learning, in particular PIP-Net, for automated diagnosis support on real-world medical imaging data. PIP-Net learns human-understandable prototypical image parts and we evaluate its accuracy and interpretability for fracture detection and skin cancer diagnosis. We find that PIP-Net's decision making process is in line with medical classification standards, while only provided with image-level class labels. Because of PIP-Net's unsupervised pretraining of prototypes, data quality problems such as undesired text in an X-ray or labelling errors can be easily identified. Additionally, we are the first to show that humans can manually correct the reasoning of PIP-Net by directly disabling undesired prototypes. We conclude that part-prototype models are promising for medical applications due to their interpretability and potential for advanced model debugging.

Authors

Christin Seifert

2 papers

Meike Nauta

1 papers

J. H. Hegeman

1 papers

TL;DR

Abstract

Authors

References42 items

PIP-Net: Patch-Based Intuitive Prototypes for Interpretable Image Classification

Reveal to Revise: An Explainable AI Life Cycle for Iterative Bias Correction of Deep Models

Explainable AI in medical imaging: An overview for clinical practitioners - Saliency-based XAI approaches.

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

Evaluating Explainable AI on a Multi-Modal Medical Imaging Task: Can Existing Algorithms Fulfill Clinical Requirements?

A ConvNet for the 2020s

Uncovering and Correcting Shortcut Learning in Machine Learning Models for Skin Cancer Diagnosis

What I Cannot Predict, I Do Not Understand: A Human-Centered Evaluation Framework for Explainability Methods

Interpretable Image Classification with Differentiable Prototypes Assignment

Understanding Radiographic Measurements Used in Foot and Ankle Surgery.

Transparency of Deep Neural Networks for Medical Image Analysis: A Review of Interpretability Methods

ProtoMIL: Multiple Instance Learning with Prototypical Parts for Whole-Slide Image Classification

An Interpretable Deep Learning Model for Covid-19 Detection With Chest X-Ray Images

Using ProtoPNet for Interpretable Alzheimer's Disease Classification

A case-based interpretable deep learning model for classification of mass lesions in digital mammography

TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentation

Neural Prototype Trees for Interpretable Fine-grained Image Recognition

AI for radiographic COVID-19 detection selects shortcuts over signal

How Useful Are the Machine-Generated Interpretations to General Users? A Human Evaluation on Guessing the Incorrectly Predicted Labels

The effect of deep convolutional neural networks on radiologists' performance in the detection of hip fractures on digital pelvic radiographs.

Radiographic analysis of adult ankle fractures using combined Danis-Weber and Lauge-Hansen classification systems

Shortcut learning in deep neural networks

Finding and removing Clever Hans: Using explanation methods to debug and improve deep models

A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis.

Interpretations are useful: penalizing explanations to align neural networks with prior knowledge

What Are the Applications and Limitations of Artificial Intelligence for Fracture Detection and Classification in Orthopaedic Trauma Imaging? A Systematic Review.

Unmasking Clever Hans predictors and assessing what machines really learn

Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)

Explanatory Interactive Machine Learning

Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

Deep learning predicts hip fracture using confounding patient and healthcare variables

Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)

This looks like that: deep learning for interpretable image recognition

MURA: Large Dataset for Abnormality Detection in Musculoskeletal Radiographs.

Right for the Right Reasons: Training Differentiable Models by Constraining their Explanations

Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning?

An Overview of Melanoma Detection in Dermoscopy Images Using Image Processing and Machine Learning

Malleolar ankle fractures. A guide to evaluation and treatment

DICOM and radiology: past, present, and future.

Recognition-by-components: a theory of human image understanding.

A Research Agenda for Hybrid Intelligence: Augmenting Human Intellect With Collaborative, Adaptive, Responsible, and Explainable Artificial Intelligence

Interpreting and Correcting Medical Image Classification with PIP-Net 19

Field of Study

Journal Information

Name

Volume