NORMA eResearch @NCI Library

Vertebra Segmentation from CT Images Using Volumetric Network (V-Net)

Mohite, Ninad (2020) Vertebra Segmentation from CT Images Using Volumetric Network (V-Net). Masters thesis, Dublin, National College of Ireland.

[img]
Preview
PDF (Master of Science)
Download (3MB) | Preview
[img]
Preview
PDF (Configuration manual)
Download (2MB) | Preview

Abstract

There was a marked rise in the number of people who suffer from vertebra disorders from past many years. For diagnosis of various vertebra or spine conditions a computer-assisted surgical systems, automatic spine or vertebra segmentation resulting from CT images is important. The spine has a complicated structure which is made up of vertebra, inter-vertebral discs, spinal cord and ribs. Hence there is a need for a robust algorithm to segment and create a model of the vertebra. In this study, V-Net is developed for segmentation of vertebra from CT images and is compared with various similar methods. It is a fully convolutional neural network(FCNN) consisting of dice loss layer and convolution tasks with max-pooling layers. The V-Net gave promising results on the limited training data using which it was trained. This method achieved soft dice loss of 0.4742 and dice coefficient of 0.5152 and the specificity and sensitivity of the model are 0.6286 and 0.5857 respectively. The experiment was done with CT images from 25 patients and it demonstrated promising results obtained from the model.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science

Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Dan English
Date Deposited: 22 Jan 2021 15:42
Last Modified: 22 Jan 2021 15:42
URI: http://norma.ncirl.ie/id/eprint/4456

Actions (login required)

View Item View Item