Home

Detection of vertebral anomalies

Image segmentation and information extraction

Project Description

This work implements a semi-automatic segmentation algorithm by region growing. Several methods have been used: *image filtering*, *morphological operations* and *image segmentation techniques*.

Introduction

Processing a computed tomography (CT) image to extract atomic measurements of vertebral height and width.

Methodology

User interaction

The algorithm asks the user to provide a starting point for the region growth. He can then click on the image that is displayed and the function returns the coordinates of the selected point.

Segmentation by region growth

Here we have the implementation of the growth algorithm. The algorithm is composed of three steps in order to work properly.

1. Preprocessing and initialization of data
2. Iterative segmentation process
3. Post-processing of segmentation.

Extraction of anatomical measurements

Once the vertebra is segmented, we measure its height and width and display them.

Here, the width of a vertebra is equal to the distance between the two most distant points horizontally, i.e. on the x-axis, while its height is equal to the distance between the two most distant points vertically, i.e. on the y-axis.

Figure: (a) original image, (b) segmentation of the vertebra, (c) extraction of the vertebra measurements.