Swarm intelligence based random walk solver for automatic segmentation of spinal cord images with voxelwise classification

Abstract

The spinal cord of humans is basically a thin long rounded structure belonging the Central Nervous System CNS that extends between the medulla oblongata and the lumbar vertebrae It acts as the important pathway for transmitting the neural signals back and forth the brain and the remaining body The evaluation of the spinal cord Magnetic Resonance MR images is basically the study of various neurological diseases it chiefly results in malfunctioning of CNS function like Multiple Sclerosis MS in which the spinal cord atrophy and also acts in the form of a measure for the assessment of the impacts of powerful neuro-protective therapies For analyzing the neurological diseases of spinal cord one of the most important steps is segmentation The recent analysis is usually carried out employing the manual segmentations with respect to the images The considerably smaller size of the spinal cord leads to manual segmentations Since the manual segmentation process analyses huge amount of data the system becomes expensive tedious and time consuming Automatic spinal cord segmentation methods have to be developed to overcome the above mentioned setbacks The newly introduced work specifying automated spinal cord segmentation is carried out using MR and Computer Tomography CT datasets Different algorithms for random walk solvers and classification methods have been utilized and their performance measured The initial contribution of the work proposed approach is performed on the basis of the automatic spinal cord segmentation with the help of MR datasets This new segmentation follows the interactive Random-Walk solvers RW along with Artificial Bee Colony ABC optimization algorithm in order to be an entirely automatic flow pipeline The initialization of the automatic segmentation pipeline is then done with a reliable voxelwise classification employing features similar to Haar and supervised machine learning technique ie Probabilistic Boosting Tree PBT along with Support Vector Machine SVM PBT SVM newline