Robust Image Registration with Absent Correspondences in Pre-operative and Follow-up Brain MRI Scans of Diffuse Glioma Patients

TL;DR

This paper introduces a three-step deep learning-based pipeline for registering pre-operative and follow-up brain MRI scans with tumors, achieving high accuracy and robustness despite missing correspondences and tissue variations.

Contribution

It presents a novel multi-level registration framework combining affine, deep learning, and non-linear optimization tailored for pathological brain MRI registration.

Findings

Achieved median error of 1.64 mm in registration accuracy.

Secured 88% success rate in the BraTS-Reg challenge.

Ranked 1st in the 2022 MICCAI BraTS-Reg challenge.

Abstract

Registration of pre-operative and follow-up brain MRI scans is challenging due to the large variation of tissue appearance and missing correspondences in tumour recurrence regions caused by tumour mass effect. Although recent deep learning-based deformable registration methods have achieved remarkable success in various medical applications, most of them are not capable of registering images with pathologies. In this paper, we propose a 3-step registration pipeline for pre-operative and follow-up brain MRI scans that consists of 1) a multi-level affine registration, 2) a conditional deep Laplacian pyramid image registration network (cLapIRN) with forward-backward consistency constraint, and 3) a non-linear instance optimization method. We apply the method to the Brain Tumor Sequence Registration (BraTS-Reg) Challenge. Our method achieves accurate and robust registration of brain MRI scans with pathologies, which achieves a median absolute error of 1.64 mm and 88\% of successful registration rate in the validation set of BraTS-Reg challenge. Our method ranks 1st place in the 2022 MICCAI BraTS-Reg challenge.