Search papers, labs, and topics across Lattice.
This study addresses the ASNR-MICCAI BraTS Local Synthesis task by developing a deterministic regression model for inpainting healthy brain tissue in T1-weighted MRIs, which is critical for providing accurate anatomical references in tumor analysis. The model employs a U-DiT architecture that integrates a downsampled global self-attention mechanism with contralateral symmetry inputs to enhance the quality of inpainting by leveraging known healthy anatomy. Results demonstrate significant improvements in distortion metrics, achieving a mean healthy-region SSIM of 0.864, PSNR of 24.7 dB, and MSE of 4.6脳10^-3 across 219 cases on the official BraTS-2026 validation leaderboard.
Leveraging contralateral symmetry in MRI inpainting leads to a remarkable boost in anatomical realism and distortion metrics, outperforming traditional methods.
The ASNR-MICCAI BraTS Local Synthesis (Inpainting) task asks for the anatomically plausible completion of healthy brain tissue within a masked region of a T1-weighted MRI, providing a tumor-free anatomical reference for downstream analysis. As the task is scored by distortion metrics (SSIM, PSNR, MSE), we build a deterministic regression model and focus on giving it inductive biases tailored to inpainting. Our network follows the U-DiT principle of performing self-attention on a downsampled token grid: a volumetric encoder-decoder imports long-range context through a downsampled global self-attention block with three-dimensional rotary position embeddings, while convolutions and skip connections preserve high-frequency detail. Two ideas drive our results. First, we constrain the attention so that occluded ("void") tokens attend only to known-healthy tokens of the same volume, with a learned bias toward each query's contralateral homologue, forcing the completion to be inferred from observed anatomy rather than from other unknown regions. Second, we add a contralateral-symmetry input that supplies the mirrored healthy hemisphere as a patient-specific prior; since the brain is approximately bilaterally symmetric and lesions are typically unilateral, this prior improves the distortion metrics at matched structural similarity. On the official BraTS-2026 validation leaderboard our submission reaches a mean healthy-region SSIM of $0.864$, PSNR of $24.7$\,dB and MSE of $4.6{\times}10^{-3}$ over $219$ cases. We further analyse the residual smoothness inherent to distortion-optimal regression and discuss its implications for anatomical realism.
