Histomorphometry is the study of the microscopic organization and structure of tissue in a quantitative manner under the microscope. This process was briefly described in a prior blog, and often involves computer-assisted technology to provide objective measures cellular architecture to aide in diagnosis – a process that is often based on subjective classification of microscopic features.
Most cancers, prostate cancer included, are diagnosed by a pathologist – who examines tissue under a microscope to determine the presence and aggressiveness of a cancer. The future of digital pathology will likely use prostate cancer quantitative histomorphometry using computer-assisted scanned image features and may serve as a new and innovative predictive tool to improve determination of aggressive phenotypes of cancer. In prior work, Dr. Robert Veltri, PhD, of the Brady Urological Institute, demonstrated that a special "spectrophotometer" microscope was accurate and capable of predicting stage, recurrence and progression of prostate cancer when examining portions of prostate cancer slides.[1]
 |
| An original, high-powered image of prostate cancer (a) is segmented into areas of individual nuclei (b), classified into individual sets (c) and segmented by the computer-alogorithm. From Ali etal [2]. |
With collaborator, Anant Madabhushi, PhD, and his team at Case Western Reserve University, Dr. Veltri is working to develop new techniques to examine the entire slide image in a high-throughput process. These novel tools could form the basis of future software tools to conduct, in cooperation with the pathologist, automated, rapid and reproducible identification and quantification of tissue histology morphologic and molecular events – enabling machine-based predictions of tumor aggressiveness and outcomes such as recurrence, metastasis and survival.
In a recent experiment, Drs. Veltri and Madabhushi examined 80 prostate cancers looking at a variety of cellular features. Through a complex computer algorithm termed adaptive active contour scheme (AdACM), they were able to distinguish features of nuclei, gland architecture, and texture and then identify the best features to discriminate Gleason grade patterns. Using these features, AdACM was able to distinguish Gleason score patterns with an accuracy of 86%.[2]
 |
| Different computer-generated features to model architecture of prostate cancer in a core of tissue from a prostate biopsy. From Ali etal [2]. |
This is just some of the exciting research ongoing at the Brady Urological Institute at Johns Hopkins that will improve the way we diagnose and treat prostate cancer.
[1] Ali S, Veltri R, Epstein JI, Christudass C, Madabhushi A.Adaptive energy selective active contour with shape priors for nuclear segmentation and gleason grading of prostate cancer.Med Image Comput Comput Assist Interv. 2011;14(Pt 1):661-9.
[2] Ali S, Veltri R, Epstein JI, Christhunesa Christudass CS, Madabhushi A. Selective Invocation of Shape Priors for Deformable Segmentation and Morphologic Classification of Prostate Cancer Tissue Microarrays. CMIG, December, 2014. dx.doi.org/10.1016/j.compmedimag.2014.11.001