Support vector machines for classification of histopathological images of brain tumour astrocytomas

Authors:
D. Glotsos;P. Spyridonos;P. Petalas;G. Nikiforidis;D. Cavouras;P. Ravazoula;P. Dadioti;I. Lekka
Affiliations:
Computer Laboratory, School of Medicine, University of Patras, Rio, Patras 265 00, Greece;Computer Laboratory, School of Medicine, University of Patras, Rio, Patras 265 00, Greece;Computer Laboratory, School of Medicine, University of Patras, Rio, Patras 265 00, Greece;Computer Laboratory, School of Medicine, University of Patras, Rio, Patras 265 00, Greece;Department of Medical Instrumentation Technology, Technological Education Institution of Athens, Ag. Spyridonos Street, Aigaleo, 122 10, Athens, greece;Department of Pathology, University Hospital, Rio, Patras, 265 00, Greece;Department of Pathology, General Anticancer Hospital METAXA, Piraeus, 18537, Greece;Department of Pathology, General Anticancer Hospital METAXA, Piraeus, 18537, Greece
Venue:
ICCMSE '03 Proceedings of the international conference on Computational methods in sciences and engineering
Year:
2003

Citing 0
Cited 3

Learning the Topological Properties of Brain Tumors

IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB)
ECM-aware cell-graph mining for bone tissue modeling and classification

Data Mining and Knowledge Discovery
Cell-graph coloring for cancerous tissue modelling and classification

Multimedia Tools and Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

A computer-based image analysis system was developed for the automatic classification of brain tumours according to their degree of malignancy using Support Vector Machines (SVMs). Morphological and textural nuclear features were quantified to encode tumour malignancy. 46 cases were used to construct the SVM classifier. Best vector was obtained performing an exhaustive search procedure in feature space. SVM classifier gave 84.8% accuracy using the leave-one-out method. To validate the systems' generalization to unseen data, 41 cases collected from a different hospital were utilized. For the validation unseen data set classification performance was 82.9%. The generalization ability of the proposed classification methodology was verified enforcing the belief that automatic characterization of brain tumours might be feasible in every day clinical routine.