Voxel-based analyses of MRI datasets are widely used for quantitative analyses in neuroscience. This technique, unfortunately, has poor statistical power and high noise level leading to a low sensitivity for detection of abnormalities. In contrast, ROI-based analyses attenuate these shortcomings by grouping the voxels corresponding to the same anatomical structure, reducing, thereby, the number of comparisons and increasing the signal-to-noise ratio. Such concerns apply to MRI-based methods used to estimate iron deposition in the brain. The sequences usually employed (multi-echo T2 and SWI) are very noisy and each technique has its own limitations. In this scenario, a neuroimaging tool that enables automatic parcellation of the entire brain and ROI-based statistical analyses is fundamental. Therefore, the use of an atlas-based approach, which performs automatically the brain parcellation, in native space, for multiple image contrasts and employs the ROI-based analyses to carry out the statistical analyses is very important and necessary to identify iron accumulation. Furthermore, atlas-based analyses handle the spatial information more efficiently, merging measures for all image contrasts and clinical data, allowing us to investigate the anatomical features for each patient. Regarding our research proposal, it would permit us to obtain for each ROI simultaneous estimates of iron deposition from multi-echo T2, volumetric reduction from T1-weighted images and white matter abnormalities from DTI images.Moreover, such approach would enable us to perform a personal assessment for each subject, and not only between-group comparisons.
News published in Agência FAPESP Newsletter about the scholarship: