Local Implicit Neural Representations for Multi-Sequence MRI Translation

TL;DR

This paper introduces a novel MRI translation method using local implicit neural representations, employing local MLPs conditioned by hypernetworks to improve the synthesis of missing MRI sequences, especially fine details and tumors.

Contribution

The paper proposes a new local implicit neural representation approach with hypernetwork-generated local MLPs for MRI sequence translation, enhancing detail recovery over classical methods.

Findings

Improved visual quality over pix2pix.

Significantly better quantitative scores (MSE and SSIM).

Local MLPs are crucial for fine detail reconstruction.

Abstract

In radiological practice, multi-sequence MRI is routinely acquired to characterize anatomy and tissue. However, due to the heterogeneity of imaging protocols and contra-indications to contrast agents, some MRI sequences, e.g. contrast-enhanced T1-weighted image (T1ce), may not be acquired. This creates difficulties for large-scale clinical studies for which heterogeneous datasets are aggregated. Modern deep learning techniques have demonstrated the capability of synthesizing missing sequences from existing sequences, through learning from an extensive multi-sequence MRI dataset. In this paper, we propose a novel MR image translation solution based on local implicit neural representations. We split the available MRI sequences into local patches and assign to each patch a local multi-layer perceptron (MLP) that represents a patch in the T1ce. The parameters of these local MLPs are generated by a hypernetwork based on image features. Experimental results and ablation studies on the BraTS challenge dataset showed that the local MLPs are critical for recovering fine image and tumor details, as they allow for local specialization that is highly important for accurate image translation. Compared to a classical pix2pix model, the proposed method demonstrated visual improvement and significantly improved quantitative scores (MSE 0.86 x 10^-3 vs. 1.02 x 10^-3 and SSIM 94.9 vs 94.3)