DISA: DIfferentiable Similarity Approximation for Universal Multimodal Registration

TL;DR

DISA introduces a differentiable approximation of similarity metrics for multimodal image registration, enabling fast, generalizable, and registration-free clinical applications across diverse modalities and anatomies.

Contribution

The paper presents a novel framework using CNN-based feature space dot-products to approximate complex similarity metrics, allowing fast, differentiable, and modality-agnostic registration.

Findings

Significantly faster than patch-based metrics

Generalizes well to unseen anatomies and modalities

Does not require registered training data

Abstract

Multimodal image registration is a challenging but essential step for numerous image-guided procedures. Most registration algorithms rely on the computation of complex, frequently non-differentiable similarity metrics to deal with the appearance discrepancy of anatomical structures between imaging modalities. Recent Machine Learning based approaches are limited to specific anatomy-modality combinations and do not generalize to new settings. We propose a generic framework for creating expressive cross-modal descriptors that enable fast deformable global registration. We achieve this by approximating existing metrics with a dot-product in the feature space of a small convolutional neural network (CNN) which is inherently differentiable can be trained without registered data. Our method is several orders of magnitude faster than local patch-based metrics and can be directly applied in clinical settings by replacing the similarity measure with the proposed one. Experiments on three different datasets demonstrate that our approach generalizes well beyond the training data, yielding a broad capture range even on unseen anatomies and modality pairs, without the need for specialized retraining. We make our training code and data publicly available.