Investigating the Role of Bilateral Symmetry for Inpainting Brain MRI

TL;DR

The paper addresses how conditioning information and hemispheric symmetry affect brain MRI inpainting. It uses a diffusion-based inpainting approach (RePaint) to systematically compare baseline versus bilateral masking, assessing impact on accuracy and variability across subcortical structures. The study finds that contralateral masking increases variability and can disrupt consistency, with the lateral ventricle and thalamus most affected, underscoring symmetry cues as influential in the model's reconstructions. These results highlight the need for careful interpretation of inpainting in clinical contexts and motivate extending to 3D models and more diverse populations.

Abstract

Inpainting has recently emerged as a valuable and interesting technology to employ in the analysis of medical imaging data, in particular brain MRI. A wide variety of methodologies for inpainting MRI have been proposed and demonstrated on tasks including anomaly detection. In this work we investigate the statistical relationship between inpainted brain structures and the amount of subject-specific conditioning information, i.e. the other areas of the image that are masked. In particular, we analyse the distribution of inpainting results when masking additional regions of the image, specifically the contra-lateral structure. This allows us to elucidate where in the brain the model is drawing information from, and in particular, what is the importance of hemispherical symmetry? Our experiments interrogate a diffusion inpainting model through analysing the inpainting of subcortical brain structures based on intensity and estimated area change. We demonstrate that some structures show a strong influence of symmetry in the conditioning of the inpainting process.

Figures (10)

• Figure 1: The schematic of the RePaint algorithm applied to a brain MRI.
• Figure 2: An example of the mean and standard deviation of the pixel intensities and $\log|\bm{J}|$ when inpainting a specific structure (Thalamus) over 10 runs on a specific subject.
• Figure 3: Plot of the mean and standard deviation of MSE (top) and $\log|\bm{J}|$ (bottom) across 20 inpainting runs on a single subject.
• Figure 4: Left: a scatter plot of inpainting mean squared error (MSE) per-run per-subject for unilateral and bilateral masking. Right: a scatter plot of the standard deviation of the pixelwise $\log|\bm{J}|$ per-run per-subject.
• Figure 5: An example of the mean and standard deviation of the pixel intensities and $\log|\bm{J}|$ when inpainting the putamen over 10 runs on one subject. The top figure is the left side, middle the right side and the bottom is both sides.