Can SPHARM-Based Features from Automated or Manually Segmented Hippocampi Distinguish Between MCI and TLE?

Abstract

Spherical Harmonics (SPHARM), when computed from hippocampus segmentations, have been shown to be useful features for discriminating patients with mild cognitive impairment (MCI) from healthy controls. In this paper we use this approach to discriminate patients with temporal lobe epilepsy (TLE) from healthy controls and for the first time, we aim to discriminate TLE patients from MCI ones. When doing so, we assess the impact of (i) using three different automated hippocampus segmentation techniques and (ii) three human raters with different qualification providing manual segmentation labels. We find that (only a fusion of) the considered automated segmentation tools deliver segmentation data which can finally be used to discriminate TLE from MCI, but for discriminating TLE from healthy controls automated techniques do not help. Further, the qualification of human segmenters has a decisive impact on the outcome of subsequent SPHARM-based classification, especially for distinguishing TLE from healthy controls (which is obviously the more difficult task).