Learning continuous-time maps improves medical image alignment accuracy
Two-Parameter Flow Map Learning for Continuous-Time Diffeomorphic Image Registration
Computer Vision and Pattern Recognition
Summary
Aligning medical images from different people helps doctors compare anatomy accurately. Most existing methods use simpler models that assume constant changes over time, which can limit accuracy. The authors introduce a new way to learn how images change over time using a two-parameter map that captures more detailed deformations without relying on slow computations during training. Their method improves alignment accuracy on brain, lung, heart, and ultrasound images while keeping the important mathematical properties of smooth deformations.
What this means in practice
- •For medical imaging teams: Improve accuracy of aligning different patients' medical images in brain, lung, cardiac MRI, and ultrasound datasets using learned continuous-time deformation maps.
- •For medical device developers: Integrate more precise image registration techniques to enhance image-guided diagnosis and treatment planning tools.
Authors
Mohammadjavad Matinkia, Nilanjan Ray
Abstract
Diffeomorphic image registration is central to medical image analysis, enabling anatomically consistent alignment across subjects. Most learning-based diffeomorphic methods model autonomous ODEs(ordinary differential equations) by parameterizing a stationary velocity field and recovering deformations via scaling-and-squaring. While non-autonomous ODEs with time-dependent velocities increase expressiveness, existing approaches rely on numerical integration to implicitly enforce flow structure that entangles model expressiveness with discretization accuracy. We propose a framework to directly learn the continuous-time solution of a non-autonomous ODE formulated as a two-parameterflow map. By enforcing cocycle consistency, a fundamental structural property of time-varying flows, we learn the flow maps without time discretization and velocity integration during training. The framework recovers diffeomorphic mappings at inference using a small number of compositions. Our proposed framework seamlessly incorporates standard registration backbones and improves alignment accuracy consistently across nine datasets while preserving diffeomorphic structure. Notably, the proposed method achieves an average Dice improvement of 2.1% on brain MRI benchmarks, a 12% TRE reduction on lung CT, and a 2.6% Dice gain on cardiac MRI and ultrasound datasets.