Peer-reviewed research in medical imaging and machine learning
Peer-reviewed research contributions in medical imaging and machine learning
An LPS-induced mouse model demonstrated that T2 MRI detects gray matter inflammation, though changes are too subtle for clinical use. This work was presented as a digital poster at the ISMRM 2022 annual meeting, showcasing our contributions to the field of neuroimaging and inflammation detection.
This work found that R2* imaging detected significant differences in the cortex of EAE mice compared to controls, while quantitative susceptibility mapping (QSM) showed no such differences. These results suggest that R2* may be more sensitive to certain pathological changes and could serve as a complementary tool to QSM analysis. This study was presented as a digital poster at the ISMRM 2023 annual meeting.
This study found increased R2*—an MRI marker of hypoxia—in the cortex of EAE mice, alongside reduced cerebral blood flow (CBF) in both inflammatory and autoimmune models, suggesting that hypoxia may be linked specifically to autoimmune-mediated damage in MS. This work was presented as an oral talk at the ISMRM 2023 annual meeting.
This study evaluated CycleGAN models—with and without spectral normalization—for generating missing MRI sequences in MS, showing improved performance with spectral normalization, especially under inconsistent scan quality. This work was presented as a traditional poster at the ISMRM 2024 annual meeting.
This study used deep learning models, including ResNet50, to predict 5-year progression in secondary progressive MS from routine MRI, achieving strong performance despite a small sample size. This work was presented as an ePoster at ECTRIMS 2024 and published in Multiple Sclerosis Journal.