Preliminary results of death cell counting based on K-mean clustering

Death cells and living cells counting after cancer drug treatment is a mandatory process for in vitro study to evaluate the effectiveness of the treatment in cancer research. The conventional process using trypan blue dye staining requires expertise and it is time-consumed and tedious work. The aim of this study was to develop a computer-assisted program that counts a number of cells by using image analysis. There were five steps to complete in this study; i) input image acquiring, ii) cell extraction from a background, iii) noise reduction, iv) cell counting and v) output with expert comparison. K-mean algorithm was selected to use to extract features and cluster objects in the images. Hough transform was also performed after completion of k-mean algorithm and noise removal. The counting results using our code had a greater number of both death cells and living cells compared with the counting results from the expert. The accuracy of death cells counting and living cells counting were in range of 33% to 97% and 74% to 100%, respectively. However, the process time was short, only 2-3 second per image. This computer-assisted program needs to further develop as a graphic user interface (GUI) to make it easier for users as well as making higher accuracy.