The impact of deep learning based image reconstruction on the quantification of DTI for assessing the severity of depression

Motivation: Previous study had reported the role of diffusion tensor imaging (DTI) in assessing severity of depression. However, DTI suffer from low image SNR which may have impact on quantification. Deep learning reconstruction (DLR) can significantly improve SNR without additional scan-time. Goal(s): Investigating the impact of DLR on DTI for evaluating severity of depression. Approach: 28 mild-to-moderate and 24 severe depression patients were involved. DTI fractional anisotropy (FA) and performance for assessing depression severity were compared between original and DLR DTI. Results: DLR DTI derived FA were smaller than that of original DTI. DLR DTI was superior to original DTI for identifying depression severity. Impact: DLR can significantly increase the SNR of DTI images which likely improve the quantification accuracy for better assessing depression severity. Therefore, the application of DLR would be beneficial for depression assessment and management.