Skip secondary menu

Semi-automatic Algorithms for Spinal Cord Volume Analysis


Human T-Lymphotropic Virus-1 Associated Myelopathy/Tropical Spastic Paraparesis (HAM/TSP) is a chronic progressive disease similar to Multiple Sclerosis (MS).  HAM/TSP is characterized by inflammation and atrophy of the spinal cord.  However, there is yet to be any robust method to quantify spinal cord atrophy as a measure of reduction in cross-sectional area from non-invasive imaging techniques.  Such a method would help to create an imaging marker for HAM/TSP and would serve as a surrogate end point in clinical trials.  This project seeks to develop an analysis pipeline for such quantification which would include semi-automatic segmentation of the whole cord and the registration of multiple contrast images.  Magnetization transfer imaging was performed on 10 MS, 10 HAM/TSP and healthy volunteers.  These images were treated in a novel manner through Analysis of Functional Neuroimages (AFNI) in an attempt to register together images of varying contrasts and to stitch together images of the cervical and thoracic spine.  AFNI algorithms successfully registered together four of five contrasts studied, and were able to calculate a transformation matrix that could be applied to thoracic images to stitch them together with cervical images.  A MATLAB program was also written to calculate cross-sectional area along the spinal cord, by a method involving perpendicular slice plane creation.  The program was tested on one MS, one HAM/TSP, and one healthy control.  The MS and HAM/TSP patients showed significant decreases in cross-sectional area as compared to the healthy control.  Nevertheless, more patients and age-and-gender-matched healthy controls need to be tested in order to produce quantitative markers for MS and HAM/TSP.  AFNI has shown much promise in spinal cord segmentation and stitching, but its registration capabilities need to be further developed in order to fully segment out the entire spinal cord.

Last updated December 14, 2012