Automatic Pulmonary Nodule Growth Measurement through CT Image Analysis based on Morphology Filtering and Statistical Region Merging

Elaheh Aghabalaei khordehchi¹, Ahmad Ayatollahi² and Mohammad Reza Daliri³

¹Biomedical Electric, Electrical Engineering Department, Iran University of Science and Technology, Tehran, Iran.

²Electrical Engineering Department, Iran University of Science and Technology, Tehran, Iran.

³Electrical Engineering Department, Iran University of Science and Technology, Tehran, Iran.

Corresponding Author E-mail: elaheh.aghabalaiee@gmail.com

Abstract: This paper proposes an innovative method for automatic detection of pulmonary nodules in Computed Tomography (CT) data and measurement of changes in the number and sizes of the detected nodules during the treatment session. In the presented method, two multi-slice CT images are first taken from the patient’s lung, each captured by a similar capturing device but at two different dates. The CT images are then analyzed and their pulmonary nodules are extracted using a novel framework based on Mathematical Morphology Filtering (MMF), Statistical Region Merging (SRM), and Support Vector Machines (SVM). The MMF step smoothes the image in order to increase its homogeneity as well as removing the noises and artifacts. The SRM algorithm segments each slice of the CT image. After connecting the boundaries of the segments in adjacent slices, three-dimensional objects are produced which are considered as nodule-candidates. These candidates are classified into nodules and non-nodules using a two-class SVM classifier. The extracted nodules in each image are then labeled and their characteristics (i.e. labels, locations, and sizes) are stored. Finally, after registering the image pair using an affine algorithm, the growth rates of the lung nodules are measured.

Keywords: Affine Algorithm; Computed Tomography; lung Nodules

[ HTML Full Text]

Back to TOC