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

详细信息    查看全文

• 作者：Dimitris Glotsos ; Panagiota Spyridonos ; Panagiota Ravazoula ; Dionisis Cavouras ; George Nikiforidis
• 刊名：Lecture Notes in Computer Science
• 出版年：2004
• 出版时间：2004
• 年：2004
• 卷：3316
• 期：1
• 全文大小：183 KB
• 刊物类别：Computer Science
• 刊物主题：Artificial Intelligence and Robotics
  Computer Communication Networks
  Software Engineering
  Data Encryption
  Database Management
  Computation by Abstract Devices
  Algorithm Analysis and Problem Complexity
  
• 出版者：Springer Berlin / Heidelberg
• ISSN：1611-3349

文摘

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.