Comparative Study of Unsupervised Segmentation Algorithms for Delineating Glioblastoma Multiforme Tumour

Glioblastoma multiforme (GBM) are fast-growing brain tumours with grade IV malignancy. They have high proliferative activity and are thus considered lethal brain tumours. Hence, the diagnosis of brain GBM tumours in early stages could aid for better treatment. Computer-aided techniques are indispens...

Full description

Saved in:
Bibliographic Details
Published in2018 3rd International Conference on Advanced Robotics and Mechatronics (ICARM) pp. 468 - 473
Main Authors Suraj, N Sri Sai Krishna, Muppalla, Vineeth, Sanghani, Parita, Ren, Hongliang
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.07.2018
Subjects
Clustering algorithms
Conferences
dice similarity coefficient
fuzzy c-means clustering
Glioblastoma Multiforme tumour
Image segmentation
k-means clustering
Mathematical model
Mechatronics
Robots
Tumors
Online AccessGet full text

Cover

More Information
Summary:Glioblastoma multiforme (GBM) are fast-growing brain tumours with grade IV malignancy. They have high proliferative activity and are thus considered lethal brain tumours. Hence, the diagnosis of brain GBM tumours in early stages could aid for better treatment. Computer-aided techniques are indispensable tools which can handle tumour segmentation with high speed and accuracy. In this paper, GBM segmentation was performed using k-means clustering (KM) and fuzzy c-means (FCM) clustering. The segmentation methods were evaluated and tested on T1 contrast-enhanced magnetic resonance image dataset of 78 patients. The performance of the two methods was compared using dice similarity coefficient (DSC). The average value of DSC for KM clustering and FCM clustering were 0.652 ± 0.105 and 0.64 ±0.121 respectively. The results obtained from this study demonstrates that KM clustering technique outperforms FCM clustering for GBM tumour segmentation.
DOI:10.1109/ICARM.2018.8610823