Increasing the Efficacy of Detecting Residual Cancer Cells Using a Whole-Slide Image Scanning Method and a Computer-Aided Detection Scheme

Abstract

1.1.Background and Objective:The purpose of this investigation is to develop and test a new computer aided detection (CAD) scheme which is able to identify the residual cancer cells from the digitalized clinical specimens for the prognostic assessment of leukemia/lymphoma. 1.2.Methods:First, a whole slide image scan was performed by a commercialized fluorescent microscopic image scanner equipped with an objective 40× lens. Then, a computerized scheme was applied to detect and segment all clinically analyzable interphase cells depicted on the scanned images, as well as to recognize and count the independent FISH-probed signal dots within each interphase cell. Five pathological specimens were used to test the performance of this new scheme. 1.3.Results:The result shows that our scheme segmented and analyzed 4546, 3807, 2880, 2240, and 849 analyzable cells in five slides of different specimens including blood, bone marrow samples respectively, among which 334, 405, 178, 117, and 24 cells are detected by the scheme as suspiciously abnormal (or residual malignant) cells. 1.4.Conclusions:Comparing to the current visual detection method, the CAD scheme identified a much larger amount of FISH-probed cells. This investigation may help more sensitively detect residual cancer cells and improve the accuracy of prognostic assessment for leukemia/lymphoma patients in the future.