Manuscript Title:

A PRECISE MODEL FOR INTERVERTEBRAL DISC LOCALIZATION AND SEGMENTATION IN MRI USING A NOVEL LU-WOA-U NET ALGORITHM

Author:

SPURTHI ADIBATTI, K.R SUDHINDRA, JOSHI MANISHA S

DOI Number:

DOI:10.17605/OSF.IO/HF7NM

Published : 2022-04-23

About the author(s)

1. SPURTHI ADIBATTI - Research Scholar, Department of ECE, BMSCE Bangalore.
2. K.R SUDHINDRA - Associate Professor, Department of ECE, BMSCE Bangalore.
3. JOSHI MANISHA S - Professor, Department of Medical Electronics, BMSCE Bangalore.

Full Text : PDF

Abstract

In the modern era, the advance technologies are entirely change the human lifestyle, behaviors and also, it is routine for daily life. For this reason, numerous diseases have been developed and cause death form illions of people every year. Although the degenerative process of the Intervertebral Disc (IVD) is not a deadly disease, this condition can’t be cured instantaneously. Therefore, to address this issue a novel U-Net based Whale Optimization Algorithm (WOA) is developed. It is one of the Machine Learning (ML) algorithms for enhancing the efficiency of the segmentation process. Moreover, MRI images are used as the dataset of proposed model. To improve the classification process the collected MRI datasets are split into multiple segments. Consequently, the proposed technique includes three main phases such as 1.DataPre-processing2. Localization and finally, 3.Segmentation. Here, redundant noise and unwanted information of the input images are neglected with the help of Gaussian filtering function. Hereafter, preprocessed images are taken as localized images, and then localized images are segmented through theU- Net segmentation model which is designed with the efficiency of Lagrangian Updated-WOA(LU-WOA). Moreover, the performance of the proposed LU-WOA-UNet model is validated through a comparative analysis over state-of- the-art models interms of accuracy, sensitivity, specificity, and DSC.Besides, the proposed model achieved significant performances over the conventional models like DCNN, SVM, NN, FCN and SOTA.


Keywords

LU-WOA, UNet, Segmentation Localization, GLCM, Multi- ModalMRI.