Software
Our lab develops software tools for quantitative OEF mapping and neuroimaging research. Tools are available for academic and research use. Please contact us at junghunc@gwu.edu if you have any questions or are interested in collaborating with our group.
Tools
A machine learning algorithm (CCTV; temporal Clustering, tissue Composition, and Total Variation) for OEF mapping that substantially improves effective signal-to-noise ratio (SNR) by clustering voxels with similar MRI signal patterns. CCTV enables robust OEF quantification and has been shown to detect OEF abnormalities in neurologic disorders.
Implementation: You can download the QQ-CCTV executable for two input formats: Dicom and NIfTTI. After downloding, please follow the instructions in Readme.txt. If you use this executable, please cite references listed in Papers section below.
A deep learning-based OEF mapping tool (NET) built on the QQ biophysics model. Using a U-Net architecture, QQ-NET achieves accurate OEF estimation with reconstruction speeds approximately 150 times faster than conventional optimization approaches, enabling practical deployment in both research and clinical settings.
Implementation: To run the code, download all the files into a folder, make sure all the libraries listed in codes/basics/unet3d_b_limit_p.py exist, and run /codes/QQ_NET_test_simul.ipynb. QQ-NET result will be saved /result/SNR100_NET_p5_trial*_overlap30.mat. To plot results in MATLAB, run /result/see_simul_horizontal.m. Please note that QQ-NET must be re-trained if the test TE sets differ from the training TE set used in the QQ-NET paper. If you use this code, please cite references listed in Papers section below.