Segmentation of Thyroid Nodules with K-Means Algorithm on Mobile Devices

Abstract

Thyroid nodules are among the most common endocrine complaints in the world. Modern imaging techniques such as ultrasound (US), computerized tomography (CT), and magnetic resonance imaging (MRI) have revealed more thyroid nodules incidentally. Therefore segmentation and quantification of them are very important for the treatment. In this paper, we have proposed a medical image processing application that can be accessed over internet by android based mobile devices. The proposed system can be thought as a decision/diagnosis support system for physicians. The aim the study is the segmentation and quantification of the crossectional areas of the thyroid nodules. The analysis results are generated and compared with the manually segmented nodule areas by using Zijdenbos Similarity Index (ZSI) on a server. The android based mobile devices are employed as clients to reach the related medical data and analysis results over internet. Obtained ZSI values have shown that proposed methodology can be used as a decision support tool for thyroid nodule evaluation with average 90% ZSI segmentation accuracy. The obtained results showed that proposed methodology will contribute to the medical image processing issues on mobile devices and help physicians to access the system via mobile devices.