A combined structural/statistical texture analysis of monolayer ovarian cancer cell nuclei

Abstract

Determining the prognosis in an early stage of human cancer can be essential for the choice of optimal therapy. Digital image analysis of cell nuclei is a very useful tool to obtain quantitative information for robust and reliable prognosis. A substantial number of papers have been published on the use of various texture analysis methods for diagnostic and prognostic work on human cancer, and most of the studies are based on texture analysis of the gray levels in the images. We will take another approach, and use a refined adaptive segmentation method developed in this thesis to describe the structures inside the cell nuclei images. The refined thresholding method is spatially adaptive within each image, while its parameters are adapted to the histogram of each image. In a novel approach, we evaluate the characteristics of the segmented structures statistically to decide the prognosis per image, and finally a rule is formed to classify each patient. The data set analyzed consists of 134 patients with early ovarian cancer. The problems with such small data sets is addressed, and a solution based on statistical bootstrapping is proposed. This gives a more robust estimate of the correct classification rate (CCR) than the traditional single CCR estimate would, and in addition gives a CCR uncertainty estimate. Dividing the data set into two groups based on DNA-ploidy - effectively introducing a simple two-step classification scheme - substantially improved the performance of the classification. Combining the structural features extracted from the objects inside each cell nucleus with the best statistical gray level feature - an adaptive entropy matrix feature from a previous study on the same material - further improved the correct classification rate, leading to a CCR close to 90\%. In conclusion, the significant improvement in correct classification rate obtained by combining the best statistical and structural texture features seems to hold a promise of very high CCRs, which would be immensely valuable in prognostic work on human cancers. This may be true beyond the present data set, and possibly quite generally. But obviously some caution is called for, and more tests on different and larger data sets should be performed.