Linear-combination (LC) modeling for metabolite level estimation
“Osprey” is a new open-source platform for linear-combination (LC) modeling for metabolite level estimation. Osprey performs all necessary steps for pre-processing, modeling and quantification of MRS data within a single open-source package for rigorous, reproducible analyses and transparent methods reporting. Osprey includes automated routines to co-register MRS voxels to segmented structural images and perform tissue-corrected quantification, and can overlay MRS voxel masks on other images (e.g. CEST, PET) to record MRS-voxel-specific image statistics. By being open-source and exposing the core modeling algorithm, Osprey enables the development of new spectral modeling approaches, such as multi-spectrum modeling, MRS fingerprinting and deep-learning-based modeling.
Osprey is currently being applied for the quantification of short-TE single-voxel data, multi-voxel/MRSI data, and multi-metabolite edited MRS data in a range of studies of neurological, psychiatric and neurodevelopmental disorders. New data formats, sequences, acquisition techniques, and features are continuously added through intensive exchange with collaborators and users.
The Osprey framework will provide the software foundation to host new data analysis methods and quantification approaches. Compared to established - often closed source - analysis software, the modularity and flexibility of the Osprey workflow will afford us to move beyond single-spectrum fitting towards more advanced modeling, including those based in deep learning across modalities.
Schematic overview of the Osprey pipeline, highlighting the modular structure of the workflow. New functions, processing methods, and modeling algorithms can be easily branched out and implemented.
The entire Osprey source code is freely available from GitHub (https://github.com/schorschinho/osprey).
The documentation can be found at https://schorschinho.github.io/osprey/.
An introduction video (OHBM 2020 software demonstration) can be found at https://www.youtube.com/watch?v=x7n8W3UCuEM (17:20 min).
Reference:
Oeltzschner G, Zöllner HJ, Hui SCN, Mikkelsen M, Saleh MG, Tapper S, Edden RAE. Osprey: Open-source processing, reconstruction & estimation of magnetic resonance spectroscopy data. J Neurosci Methods 2020;343:108827.