Automated brain atrophy quantification and evaluation using spatial resolution enhancement.

Brain atrophy is one of the most common features of neurodegenerative diseases and is particularly critical in the early diagnosis of conditions like Alzheimer's and multiple sclerosis. Automated segmentation and quantification are highly desirable in brain atrophy evaluation but existing methods require high-quality MRI scans with isotropic resolution. However in practice, clinicians usually choose to reduce the number of slices to save time, and because of their anisotropic resolution, the morphometric analysis cannot be implemented. Here we propose the use of an inter-slice interpolation network to increase spatial resolution to isotropic to accomplish morphometric analysis of brain atrophy, which is not feasible to perform with clinical scans. To achieve fully automated quantification, we set up a series of indicators that cover multiple aspects such as holistic and localized, structural and cortical, based on our clinical experience. As a result, a fully automated diagnostic process for pathological brain atrophy is completed. Our results on the IXI and ADNI datasets show that the slice spacing method produces realistic and reliable anatomy and achieves high diagnostic accuracy, and our indicators are well suited to clinical application scenarios.