Automatic Positioning of Hippocampus Deformable Mesh Models in Brain MR Images Using a Weighted 3D-SIFT Technique

Author(s):Show less -

Korb, Matheus Muller ; Ferrari, Ricardo Jose ; Alzheimer's Dis Neuroimaging ; Gervasi, O ; Murgante, B ; Misra, S ; Garau, C ; Blecic, I ; Taniar, D ; Apduhan, BO ; Rocha, AMAC ; Tarantino, E ; Torre, CM ; Karaca, Y

• Document type: Journal article
• Source: COMPUTATIONAL SCIENCE AND ITS APPLICATIONS - ICCSA 2020, PT II; v. 12250, p. 16-pg., 2020-01-01.

Abstract

Automatic hippocampus segmentationin Magnetic Resonance (MR) images is an essential step in systems for early diagnostic and monitoring treatment of Alzheimer's disease (AD). It allows quantification of the hippocampi volume and assessment of their progressive shrinkage, considered as the hallmark symptom of AD. Among several methods published in the literature for hippocampus segmentation, those using anatomical atlases and deformable mesh models are the most promising ones. Although these techniques are convenient ways to embed the shape of the models in the segmentation process, their success greatly depend on the initial positioning of the models. In this work, we propose a new keypoint deformable registration technique that uses a modification of the 3D Scale-Invariant Feature Transform (3D-SIFT) and a keypoint weighting strategy for automatic positioning of hippocampus deformable meshes in brain MR images. Using the Mann-Whitney U test to assess the results statistically, our method showed an average improvement of 11% over the exclusive use of Affine transformation, 30% over the original 3D-SIFT and 7% over the non-weighted point procedure. (AU)

• FAPESP's process: 18/08826-9 - Development of feature engineering and deep learning techniques applied to the classification of magnetic resonance images in healthy cognitive aging, mild cognitive impairment and Alzheimer's Disease
• Grantee: Ricardo José Ferrari
• Support Opportunities: Regular Research Grants