Fast learning-based registration of sparse 3D clinical images

Abstract

We introduce SparseVM, a method to register clinical 3D MR scans faster and more accurately than previously possible. Deformable alignment, or registration, of clinical scans is a fundamental task for many medical image applications such as longitudinal population studies. Most registration algorithms are designed for high-resolution research-quality scans, and under-perform when applied to clinical data. Clinical scans present unique challenges because, in contrast to research-quality scans, clinical scans are often sparse, missing up to 85% of the slices available in research scans. We build on a state-of-the-art learning-based registration method to improve the accuracy of sparse clinical image registration. We evaluate our method on both a clinically-acquired MRI dataset of stroke patients, and on a simulated sparse clinical scan dataset. SparseVM registers MR scans in under a second on a GPU, which is over 1000 x faster than the most accurate clinical registration methods, without compromising accuracy. SparseVM has statistically significant improvements in runtimes over all baselines, and a statistically significant improvement in accuracy over the fastest baseline. Because of these contributions, SparseVM enables clinical analyses that were not previously possible. Our code is publicly available at voxelmorph. mit . edu.