Osteosarcoma segmentation in MRI using dynamic Harmony Search based clustering

Abstract

In this paper, the automatic segmentation of Osteosar-coma in MRI images is formed as a clustering problem. Subsequently, a new dynamic clustering algorithm based on the Harmony Search (HS) hybridized with Fuzzy C-means (FCM) called DCHS is proposed to automatically segment the Osteosarcoma MRI images in an intelligent manner. The concept of variable length in each harmony memory vector is applied to encode variable numbers of candidate cluster centers at each iteration. Furthermore, a new HS operator, called the 'empty operator' is introduced to support the selection of empty decision variables in the harmony memory vector. FCM is incorporated in DCHS to fine tune the segmentation results. Our approach uses multi-spectral information from STIR (Short Tau Inversion Recovery) and T2-weighted MRI sequences. We used a subset of Haralick texture features and pixel intensity values as a feature space to DCHS to delineate the tumour volume. The segmentation results were statistically evaluated against manually delineated data for four patients. Promising results were obtained with average of 0.72 of Dice measurement. In addition, we also propose a method to identify necrotic tissue within the tumour in order to monitor drug-induced necrosis of tumor tissue.