On the Classification of Prostate Pathological Images Based on Gleason Score

Prostate cancer is one of the most frequent cancers caused in men and automated classification results which can be provided as objective references are of great significance. Here we present a study of classification of histological images of prostate based on both morphological features and textural features. At first we get two tissues of prostate cancer which including nuclei, lumen from the image, then a total of 40 morphological and textural features from each digitized images of histological prostate tissue specimens. After feature selection, a support vector machine (SVM) is used to classify the digitized histology slides into two classes: "Gleason score=7". In our experiments the SVM classifier achieving an accuracy of 90.67% within the training set and 74.81% within the test set, respectively.

[1]  Anant Madabhushi,et al.  AUTOMATED GRADING OF PROSTATE CANCER USING ARCHITECTURAL AND TEXTURAL IMAGE FEATURES , 2007, 2007 4th IEEE International Symposium on Biomedical Imaging: From Nano to Macro.

[2]  Ehud Rivlin,et al.  System for Computer-Aided Multiresolution Microscopic Pathology Diagnostics , 2006, Fourth IEEE International Conference on Computer Vision Systems (ICVS'06).

[3]  Mikhail Teverovskiy,et al.  Multifeature Prostate Cancer Diagnosis and Gleason Grading of Histological Images , 2007, IEEE Transactions on Medical Imaging.

[4]  Luo Shu-qian.  Application of support vector machine in medical image segmentation , 2007 .

[5]  Franz Schweiggert,et al.  On the Classification of Prostate Carcinoma With Methods from Spatial Statistics , 2007, IEEE Transactions on Information Technology in Biomedicine.