Erase to Enhance: Data-Efficient Machine Unlearning in MRI Reconstruction

TL;DR

This paper demonstrates that machine unlearning can be effectively applied to MRI reconstruction tasks, enabling removal of unwanted data or biases without full retraining, and maintaining high image quality with limited data.

Contribution

It introduces a protocol for applying machine unlearning in MRI reconstruction, showing its feasibility and benefits for bias removal and data privacy in medical imaging.

Findings

Unlearning can remove hallucinations in MRI reconstructions.

High performance is maintained with only a subset of retained data.

Unlearning does not require full retraining of models.

Abstract

Machine unlearning is a promising paradigm for removing unwanted data samples from a trained model, towards ensuring compliance with privacy regulations and limiting harmful biases. Although unlearning has been shown in, e.g., classification and recommendation systems, its potential in medical image-to-image translation, specifically in image recon-struction, has not been thoroughly investigated. This paper shows that machine unlearning is possible in MRI tasks and has the potential to benefit for bias removal. We set up a protocol to study how much shared knowledge exists between datasets of different organs, allowing us to effectively quantify the effect of unlearning. Our study reveals that combining training data can lead to hallucinations and reduced image quality in the reconstructed data. We use unlearning to remove hallucinations as a proxy exemplar of undesired data removal. Indeed, we show that machine unlearning is possible without full retraining. Furthermore, our observations indicate that maintaining high performance is feasible even when using only a subset of retain data. We have made our code publicly accessible.