SBIR-STTR Award

The purpose of this research is to establish the feasibility of utilizing color computer vision and neural network technologies to automate the white cell differential procedure currently performed by medical technologists. Automation of the blood differential procedure is needed to improve the accuracy and consistency of this procedure in the detection of leukemia, infection, and other disorders and to monitor the effects of cancer chemotherapy and radiation treatment. The relatively high error rates experienced by medical technologists in performing this procedure are associated with (1) the tedium of extended viewing through a microscope, (2) the difficulty of identifying often subtle differences in cell characteristics, and (3) human deficiencies (e.g., mood, training). A growing shortage of medical technologists, which results in delayed test results and escalating labor costs, is another reason for automation of the blood analysis procedure.Phase II will involve development of a prototype system with the capability to perform the entire microscopic blood differential procedure (WBC differential, RBC morphology, platelet estimation) for under $100,000.Awardee's statement of the potential commercial applications of the research:Success in automating the blood differential procedure will establish the technological basis for fully or partially automating the numerous microscopic cell analysis procedures carried out in various medical disciplines. All of these procedures are subject to problems of inaccuracy, delayed test turnaround, and escalating labor costs. An automated system that overcomes these problems has tremendous commercial potential.National Cancer Institute (NCI)

Phase I demonstrated the feasibility of using Color Computer Vision and Neural Network technologies (hereafter referred to as "Neural Technologies") in classifying white blood cells. Phase II seeks to develop an instrument that fully automates the microscopic differential count using Neural Technologies. Despite the wide use of flow cytometry differential counters, there are in excess of 100 million microscopic differential counts performed annually in the United States. An instrument that automates this procedure is needed to improve the accuracy and consistency in the diagnosis and monitoring of abnormal leukocytes. The accurate and consistent classification of abnormal white cells (i. e. myelocytes, variant lymphocytes, blasts, etc.) will greatly improve the ability to monitor patients' progress during treatment (chemotherapy, radiation, etc.). Automation of the leukocyte differential is also needed to help counter the growing shortage of Technologists and the associated delays in test reporting. The automation of other microscopic procedures not involving blood cells (PAP smears, tissue sections, etc.) is possible using Neural Technologies and will be the subject of subsequent investigations.Awardee's statement of the potential commercial applications of the research:The successful automation of the microscopic differential count will establish the technological basis and commercial potential for the automation of many other microscopic cell analysis procedures (bone marrow, PAP smear, tissue sections, etc.). All of these procedures are subject to the problems of inaccuracy, inconsistency, delayed test turnaround, and escalating labor costs. An instrument that helps overcome these problems has tremendous commercial potential.National Cancer Institute (NCI)