Multimodal Deformable Image Registration for Long-COVID Analysis Based on Progressive Alignment and Multi-perspective Loss

TL;DR

This work proposed an end-to-end multimodal deformable image registration method that achieves superior performance for aligning long-COVID lung CT and proton density MRI (pMRI) data, and incorporates a novel Multi-perspective Loss (MPL) function, enhancing state-of-the-art deep learning methods for monomodal registration by making them adaptable for multimodal tasks.

Abstract

Long COVID is characterized by persistent symptoms, particularly pulmonary impairment, which necessitates advanced imaging for accurate diagnosis. Hyperpolarised Xenon-129 MRI (XeMRI) offers a promising avenue by visualising lung ventilation, perfusion, as well as gas transfer. Integrating functional data from XeMRI with structural data from Computed Tomography (CT) is crucial for comprehensive analysis and effective treatment strategies in long COVID, requiring precise data alignment from those complementary imaging modalities. To this end, CT-MRI registration is an essential intermediate step, given the significant challenges posed by the direct alignment of CT and Xe-MRI. Therefore, we proposed an end-to-end multimodal deformable image registration method that achieves superior performance for aligning long-COVID lung CT and proton density MRI (pMRI) data. Moreover, our method incorporates a novel Multi-perspective Loss (MPL) function, enhancing state-of-the-art deep learning methods for monomodal registration by making them adaptable for multimodal tasks. The registration results achieve a Dice coefficient score of 0.913, indicating a substantial improvement over the state-of-the-art multimodal image registration techniques. Since the XeMRI and pMRI images are acquired in the same sessions and can be roughly aligned, our results facilitate subsequent registration between XeMRI and CT, thereby potentially enhancing clinical decision-making for long COVID management.

Figures (3)

• Figure 1: Illustration of our proposed architecture: The network initially predicts the affine transformation $\phi_{\text{affine}}$. Subsequently, cascaded networks iteratively compute the displacement fields. At each cascade $n$, the network predicts the displacement field $\phi_{\text{n}}$ using the input image $M^{\text{(n)}}$ warped by the displacement field $\phi_{\text{n-1}}$ from the previous cascade $n-1$.
• Figure 2: Visualisation of CT-pMRI registration through a wide range of methods: The top row reflects the input pMRI image before registration and the results from state-of-the-art methods, whereas the bottom row shows the fixed image (CT) overlaid by edges extracted from pMRI image (in green). Red arrows point out areas of mis-registration.
• Figure 3: Visualization of the CT-XeMRI results before and after registration.