Automated diagnosis of brain tumours using a novel density estimation method for image segmentation and independent component analysis combined with support vector machines for image classification

Glotsos, D; Spyridonos, P; Ravazoula, P; Cavouras, D; Nikiforidis, G

dc.contributor.author	Glotsos, D	en
dc.contributor.author	Spyridonos, P	en
dc.contributor.author	Ravazoula, P	en
dc.contributor.author	Cavouras, D	en
dc.contributor.author	Nikiforidis, G	en
dc.date.accessioned	2014-03-01T01:53:38Z
dc.date.available	2014-03-01T01:53:38Z
dc.date.issued	2004	en
dc.identifier.issn	03029743	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/27086
dc.relation.uri	http://www.scopus.com/inward/record.url?eid=2-s2.0-35048864910&partnerID=40&md5=a049638ab4cae33cf1fe0b3311f95cc3	en
dc.title	Automated diagnosis of brain tumours using a novel density estimation method for image segmentation and independent component analysis combined with support vector machines for image classification	en
heal.type	journalArticle	en
heal.publicationDate	2004	en
heal.abstract	A computer-aided system was developed for the automatic diagnosis of brain tumours using a novel density estimation method for image segmentation and independent component analysis (ICA) combined with Support Vector Machines (SVM) for image classification. Images from 87 tumor biopsies were digitized and classified into low and high-grade. Segmentation was performed utilizing a density estimation clustering method that isolated nuclei from background. Nuclear features were quantified to encode tumour malignancy. 46 cases were used to construct the SVM classifier. ICA determined the most important feature combination. Classifier performance was evaluated using the leave-one-out method. 41 cases collected from a different hospital were used to validate the systems' generalization. For the training set the SVM classifier gave 84.9%. For the validation set classification performance was 82.9%. The proposed methodology is a dynamic new alternative to computer-aided diagnosis of brain tumours malignancy since it combines robust segmentation and high effective classification algorithm. © Springer-Verlag Berlin Heidelberg 2004.	en
heal.journalName	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)	en
dc.identifier.volume	3316	en
dc.identifier.spage	1058	en
dc.identifier.epage	1063	en