Preliminary Results from the 2024 MRSI Data Processing and Quantification Challenge

Motivation: Realistic simulated training/testing MRSI data based on in vivo data are necessary for development of nuisance signal removal and spectral analysis methods. Goal(s): To test data processing/analysis pipelines' ability to create accurate metabolite maps from simulated FID-MRSI datasets corrupted by noise/nuisance signals. Approach: Challenge: 1) Nuisance signal removal and spectral quantification on data contaminated by noise, baseline signals, spectral distortions, and residual water/lipid, and 2) Spectral quantification on data contaminated by noise, baseline signals, and spectral distortions. Results: Nine teams signed up, three submitted: each challenge received two submissions. Variability in results were seen. Data and code are available freely to the community. Impact: This is a resource, both code and data, that the MRS community can use to create repeatable simulated data based on real world data inputs. It is a platform that encourages collaboration to simplify creation of reusable infrastructure.