# PRISM: Differentiable Analysis-by-Synthesis for Fixel Recovery in Diffusion MRI

Source: [PRISM: Differentiable Analysis-by-Synthesis for Fixel Recovery in Diffusion MRI](<https://www.sota2.com/research/paper/69cdf7f50636e5bf4afcc9cd>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-03T05:00:04.568Z

Authors: Mohamed Abouagour, Atharva Shah, Eleftherios Garyfallidis

Published: 2026-03-31T21:22:49.000Z

## Abstract

Diffusion MRI microstructure fitting is nonconvex and often performed voxelwise, which limits fiber peak recovery in narrow crossings. This work introduces PRISM, a differentiable analysis-by-synthesis framework that fits an explicit multi-compartment forward model end-to-end over spatial patches. The model combines cerebrospinal fluid (CSF), gray matter, up to K white-matter fiber compartments (stick-and-zeppelin), and a restricted compartment, with explicit fiber directions and soft model selection via repulsion and sparsity priors. PRISM supports a fast MSE objective and a Rician negative log-likelihood (NLL) that jointly learns sigma without oracle information. A lightweight nuisance calibration module (smooth bias field and per-measurement scale/offset) is included for robustness and regularized to identity in clean-data tests. On synthetic crossing-fiber data (SNR=30; five methods, 16 crossing angles), PRISM achieves 3.5 degrees best-match angular error with 95% recall, which is 1.9x lower than the best baseline (MSMT-CSD, 6.8 degrees, 83% recall); in NLL mode with learned sigma, error drops to 2.3 degrees with 99% recall, resolving crossings down to 20 degrees. On the DiSCo1 phantom (NLL mode), PRISM improves connectivity correlation over CSD baselines at all four tracking angles (best r=.934 at 25 degrees vs. .920 for MSMT-CSD). Whole-brain HCP fitting (~741k voxels, MSE mode) completes in ~12 min on a single GPU with near-identical results across random seeds.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Fiber recovery on 16-angle synthetic crossing-fiber benchmark SNR=30](<https://www.sota2.com/research/sota/fiber-recovery-on-16-angle-synthetic-crossing-fiber-benchmark-snr-30.md>): Fiber recovery; 16-angle synthetic crossing-fiber benchmark SNR=30; Angular Error (1-fib): 0.7.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.00250>)

[PDF](<https://arxiv.org/pdf/2604.00250.pdf>)
