Background: Analysis of chromatin texture may improve both the diagnosis and the assessment of the prognosis of prostate cancer. Confocal laser scanning microscopy (CLSM) allows performing measurements in nuclei reconstructed in 3-D. The aim of this study was to evaluate the clinical usefulness of 3-D texture analysis of prostate tissue.
Methods: Image stacks of eight prostate cancer sections were obtained by CLSM of both benign and malignant areas. Texture feature values were computed for individual nuclei. The discriminative power of the texture features was established by receiver operating characteristic (ROC) analysis and linear discriminant analysis (LDA).
Results: Texture features were identified that could discriminate between benign and malignant nuclei. LDA correctly classified 89% of the nuclei of the pooled set of benign and malignant nuclei.
Conclusions: 3-D nuclear texture features allow discrimination of most benign and malignant prostate nuclei. We estimate that the classification rates can be increased by improving the image quality.
(c) 2006 Wiley-Liss, Inc.