An improved unsupervised mapping technique using AMSOM for neurodegenerative disease detection Online publication date: Wed, 25-Apr-2018
by Isha Suwalka; Navneet Agrawal
International Journal of Computational Systems Engineering (IJCSYSE), Vol. 4, No. 2/3, 2018
Abstract: The most challenging aspect in medical imaging is the accuracy of detection of neurodegenerative diseases. The advent of new imaging techniques has yet limited manual evaluations, manual reorientation and other time consuming limitations with reduced resolution. Therefore, there is a need to develop efficient algorithm for proper detection with quantitative information of significance for the clinicians. The proposed algorithm includes improved adaptive moving self organising mapping (AMSOM) which trains the extracted features along with mini-mental state examination (MMSE) factor and volumetric parameter using volume-based method (VBM) for computing feature dataset which in total improves time iteration rate, mean square error, sensitivity and accuracy. The algorithm is an improved version of moving mapping method which on one hand tackles drawback of SOM of fixed grid mapping and improves neighbourhood function of neuron which provides better detection and classification yielding promising results. It further improves performance of AMSOM by better visualisation of the input dataset and provides a framework for determining the efficient parameters. This paper uses real MRI dataset taken from OASIS having a cross-sectional collection of 416 subjects aged 18 to 96. The analysis includes different comparison of mapping approaches that reveals features associated to the Alzheimer disease.
Online publication date: Wed, 25-Apr-2018
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Computational Systems Engineering (IJCSYSE):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.
If you still need assistance, please email firstname.lastname@example.org