x
Overview
Our semi automated pipeline uses Freesurfer to reconstruct the structural T1 image. The accuracy of Freesurfer parcellation and segmentation are reported to be comparable with manual labeling. The reconstructed T1 image will go through two independent and iterative level of manual correction and quality control. In the first level a single trained technician will go through the correction steps by editing brain and white-matter masks then adding as many control points as possible to achieve the satisfactory results. This is an iterative process and iteration continues until the results are satisfactory. In the second level, another trained operator checks the quality of the cortical and subcortical segmentation by overlaying the segmented regions boarder on top of the T1 image. Any discrepancy is reported back to the first technician for further correction. This iterative process continues till the satisfactory results are reached. Both operators are intentionally kept blind to the subject’s demographic and health situation. FreeSurfer test-retest reliability on repeated scan of the same 43 subjects were 94.69% and the detailed ROI based measure are given in the following box-plots.
However after manual correcting both set of repeated scan independently, the test-retest reliability increased to 99.72%. The detailed ROI based measure of differences are given in the following box-plots.
The intra-rater reliability of the manual correction process for 50 randomly selected scans were 99.73% and the detailed ROI based difference measures are given in the following box-plots.
Finally, the inter-rater reliability of the manual correction process for the same randomly selected 50 subjects were 99.68%. and the detailed ROI based difference measure are given in the following box-plots.
Related Publication
Q. R. Razlighi, H. Oh, D. O’Shea, Y. Gazes, D. B. Parker, S. Lee, C. Habeck, and Y. Stern , “Dynamic patterns of brain structure-behavior correlation across the lifespan”, Cerebral Cortex, In Press, Jul. 2016.
J. Steffener, D. Barulli, C. Habeck, D. O’Shea, Q. R. Razlighi, Y. Stern. “The role of education and verbal abilities in altering the effect of age-related gray matter differences on cognition”. PLoS One. vol. 9, no 3, 2014.
S. Blum, C. Habeck, J. Steffener, Q. R. Razlighi, and Y. Stern, “ Functional connectivity of the posterior hippocampus is more dominant as we age,” Cognitive Neuroscience , vol. 5, no. 3, pp. 150-159, 2014.
Software Released
Not applicable.