Intelligent Microscope Workstation for Cancer Diagnosis

Preston, Kendall

Abstract

This SBIR Phase I proposal is submitted by the Pathology Imaging Corp. (PIC) in the spirit of the National Cancer Act of 1971 which gave the Dept. of Health and Human Services (then Health, Education, and Welfare) the responsibility to """"""""speed the translation of research results into widespread application."""""""" During 1984-1990 academia (Carnegie Mellon University and the University of Southern California) using the $180,000 Coulter Biomedical Research robot microscope, the diff3/50, interfaced to a $80,000 Interdata model 3230 computer, conducted research under a series of NIH grants cumulatively funded at over one million dollars which achieved intelligent performance in subtyping follicular lymphomas. The resulting system, when teamed with the expert visual skills of the pathologist in locating follicles, achieved performance far exceeding that of either the robot or the pathologist acting alone. At the same time PIC funded the reduction to practice of a proprietory $5,000 computerized camera for digital microscopy having a network interface to personal computers. Using this camera and PC-based image analysis software by Kensal Consulting (Tucson) in a joint project with the Department of Pathology, University of Texas, PIC reproduced IN ONE DAY the results achieved by CMU/USC. Since the patient sample was too small to draw statistical conclusions, it is therefore the purpose of this proposal to request funding to conduct a definitive study of 100-200 patients (five times that of the CMU/USC study) and simultaneously determine the robustness of the algorithmic environment with regard to resolution, magnification, color, etc.
The aim of this Phase I research would be to reduce to practice and exhaustively test the CMU/USC cancer diagnosis which could be manufactured commercially and sold for widespread use nationwide during Phase III at a price of $15,000 or less per unit.