MRI Thigh Sequences in Determining the Tumor Size Using Fuzzy C-Means for Patients with Osteosarcoma

Abstract

Osteosarcoma is the common type of bone cancer in children and adolescents. A magnetic resonance imaging (MRI) is one of the medical imaging techniques used by specialist to diagnose the medical conditions of Osteosarcoma patient. A radiofrequency pulse and gradient sequence known as MRI sequence produces a set of pictures with a specific appearance. In clinical, radiologists need to interpret MRI images and correlating them from various sequences for medical image findings. The process requires a lot of human input and therefore it is subjective, time-consuming, and non-reproducible. Image segmentation can be used to automate MR images into different segments. In image processing, various algorithms used to segment the medical images into region. However, due to the overlap of grayscale pixel values make the segmentation process becomes very difficult and challenging. The purpose of this study is to extract tumor on MRI Osteosarcoma based on three MRI thigh sequences namely T1, T2 and T1_FSE+GADO. The area and perimeter of the extracted tumor are then compared with the ground truth. In this study, two algorithms namely OTSU Thresholding (OT) and Fuzzy C-Means (FCM) were used to perform the segmentation on the MRI Osteosarcoma images. The performance of these two algorithms on segmenting the MRI Osteosarcoma from three MRI sequences are compared and discuss. The result shows that FCM could discriminate the abnormal region better in T1_FSE+GADO sequence. The average percentage error for area in T1_FSE+GADO sequence is 6.20% and average percentage error for perimeter is 6.74% compared to T2 sequence which is 7.18% and 7.71%.