Fully Automated Hippocampus Segmentation using T2-informed Deep Convolutional Neural Networks.

Hippocampal atrophy (tissue loss) has become a fundamental outcome parameter in clinical trials on Alzheimer's disease. To accurately estimate hippocampus volume and track its volume loss, a robust and reliable segmentation is essential. Manual hippocampus segmentation is considered the gold standard but is extensive, time-consuming, and prone to rater bias. Therefore, it is often replaced by automated programs like FreeSurfer, one of the most commonly used tools in clinical research. Recently, deep learning-based methods have also been successfully applied to hippocampus segmentation. The basis of all approaches are clinically used T1-weighted whole-brain MR images with approximately 1 mm isotropic resolution. However, such T1 images show low contrast-to-noise ratios (CNRs), particularly for many hippocampal substructures, limiting delineation reliability. To overcome these limitations, high-resolution T2-weighted scans are suggested for better visualization and delineation, as they show higher CNRs and usually allow for higher resolutions. Unfortunately, such time-consuming T2-weighted sequences are not feasible in a clinical routine. We propose an automated hippocampus segmentation pipeline leveraging deep learning with T2-weighted MR images for enhanced hippocampus segmentation of clinical T1-weighted images based on a series of 3D convolutional neural networks and a specifically acquired multi-contrast dataset. This dataset consists of corresponding pairs of T1- and high-resolution T2-weighted images, with the T2 images only used to create more accurate manual ground truth annotations and to train the segmentation network. The T2-based ground truth labels were also used to evaluate all experiments by comparing the masks visually and by various quantitative measures. We compared our approach with four established state-of-the-art hippocampus segmentation algorithms (FreeSurfer, ASHS, HippoDeep, HippMapp3r) and demonstrated a superior segmentation performance. Moreover, we found that the automated segmentation of T1-weighted images benefits from the T2-based ground truth data. In conclusion, this work showed the beneficial use of high-resolution, T2-based ground truth data for training an automated, deep learning-based hippocampus segmentation and provides the basis for a reliable estimation of hippocampal atrophy in clinical studies.

Cross-sectional and Longitudinal Assessment of Brain Iron Level in Alzheimer Disease Using 3-T MRI.

Background Deep gray matter structures in patients with Alzheimer disease (AD) contain higher brain iron concentrations. However, few studies have included neocortical areas, which are challenging to assess with MRI. Purpose To investigate baseline and change in brain iron levels using MRI at 3 T with R2* relaxation rate mapping in individuals with AD compared with healthy control (HC) participants. Materials and Methods In this prospective study, participants with AD recruited between 2010 and 2016 and age-matched HC participants selected from 2010 to 2014 were evaluated. Of 100 participants with AD, 56 underwent subsequent neuropsychological testing and brain MRI at a mean follow-up of 17 months. All participants underwent 3-T MRI, including R2* mapping corrected for macroscopic B0 field inhomogeneities. Anatomic structures were segmented, and median R2* values were calculated in the neocortex and cortical lobes, basal ganglia (BG), hippocampi, and thalami. Multivariable linear regression analysis was applied to study the difference in R2* levels between groups and the association between longitudinal changes in R2* values and cognition in the AD group. Results A total of 100 participants with AD (mean age, 73 years ± 9 [standard deviation]; 58 women) and 100 age-matched HC participants (mean age, 73 years ± 9; 60 women) were evaluated. Median R2* levels were higher in the AD group than in the HC group in the BG (HC, 29.0 sec-1; AD, 30.2 sec-1; P = .01) and total neocortex (HC, 17.0 sec-1; AD, 17.4 sec-1; P < .001) and regionally in the occipital (HC, 19.6 sec-1; AD, 20.2 sec-1; P = .007) and temporal (HC, 16.4 sec-1; AD, 18.1 sec-1; P < .001) lobes. R2* values in the temporal lobe were associated with longitudinal changes in Consortium to Establish a Registry for Alzheimer's Disease total score (ß = -3.23 score/sec-1, P = .003) in participants with AD independent of longitudinal changes in brain volume. Conclusion Iron concentration in the deep gray matter and neocortical regions was higher in patients with Alzheimer disease than in healthy control participants. Change in iron levels over time in the temporal lobe was associated with cognitive decline in individuals with Alzheimer disease. © RSNA, 2020 Online supplemental material is available for this article.