A Medical Device Daily

Health Discovery (Savannah) said the U.S. Patent and Trademark Office has granted patent No. 7,383,237 on its claims covering the use of Support Vector Machines (SVMs) in computer-aided image analysis for analysis of digitized microscopic images of medical specimens. This new patent is a continuation of the company's patent No. 6,996,549, which covers a broad range of uses of SVMs for image analysis.

The claims of Health Discovery's newest patent focus on a method and computer system for analyzing medical images generated during microscopic evaluation of cytology specimens and tissue samples. Examples of commercial diagnostic and prognostic applications of the patented technology include circulating cell screening and cytopathology slide analysis, among others.

Circulating tumor cells and/or circulating epithelial cells in the blood of cancer patients have proven to be a strong, independent predictor of overall survival and progression free survival, the company noted. Health Discovery said these cells also could be used for serial monitoring of patients for improved clinical management of metastatic cancer.

Circulating cell analysis involves the generation of large numbers of microscopic "snapshots" of a sample chamber containing the patient's blood. Currently, a small number of these snapshots are then selected for visual examination by a technologist or pathologist, the company said. However, SVM-aided image analysis using the patented method could permit automated and rapid analysis of all of the sample images, greatly increasing the sensitivity and accuracy of such tests.

Cancer screening using cervical cytology slides, or, Pap smear testing, and pathological evaluation of biopsied tissue are important applications of the patented technology which are the subject of the recently-announced license, development and commercialization agreement between Health Discovery and DCL Medical Laboratories (Indianapolis), which includes a license under Health Discovery's newest patent. The SVM-based imaging method is expected to improve the sensitivity of detection for endometrial and cervical cancers and significantly improve the specificity of ovarian cancer diagnosis, according to the company. The new CAD-based digital pathology SVM-based algorithms could offer a faster, highly objective, and more accurate interpretation of cells to assist pathologists in making orrect diagnoses for physicians and their patients.

"This new patent goes to the heart of [Health Discovery's] vision of redefining the relationships between diagnostics and treatments for personalized medicine using our state-of-the-art SVM technology," said CEO/Chairman Stephen Barnhill, MD.

Barnhill said the U.S. cytology market was estimated to be about $2 billion in 2006 and continues to grow "at a rapid pace." He said Health Discovery's SVM-based image analysis provides a "valuable" tool for every facet of this market, assisting pathologists in clinical laboratories, hospitals, academic centers and medical/pharmaceutical industries by providing "an accurate, objective diagnostic interpretation to doctors for their patients.

"We expect the technology to have enormous value as a 'second opinion,' so much so that insurers might even make the analysis a mandatory part of every cytopathology test," Barnhill said.

With the issuance of this new patent, Health Discovery now holds the exclusive rights to 31 issued U.S. and foreign patents covering uses of SVM and fractal genomics modeling technology for discovery of knowledge from large data sets. Other patents issued to the company cover methods and systems for pre-processing of data to enhance knowledge discovery using SVM's, analysis of data using multiple support vector machines and for multiple data sets, and providing SVM analysis services over the Internet. Health Discovery's pending U.S. and foreign patent applications cover numerous improvements to, and applications of, SVMs including computer-aided image analysis using SVMs, methods of feature selection for enhanced SVM efficiency and biomarkers for colon cancer, prostate cancer, BPH and renal cancer discovered with these methods, and the use of SVM's for analysis of spectral data, such as mass spectrometry data used for protein analysis.