RESUMEN
Cell segmentation and counting is often required in disciplines such as biological research and medical diagnosis. Manual counting, although still employed, suffers from being time consuming and sometimes unreliable. As a result, several automated cell segmentation and counting methods have been developed. A main component of automated cell counting algorithms is the image segmentation technique employed. Several such techniques were investigated and implemented in the present study. The segmentation and counting was performed on antibody stained brain tissue sections that were magnified by a factor of 40. Commonly used methods such as the circular Hough transform and watershed segmentation were analysed. These tests were found to over-segment and therefore over-count samples. Consequently, a novel cell segmentation and counting algorithm was developed and employed. The algorithm was found to be in almost perfect agreement with the average of four manual counters, with an intraclass correlation coefficient (ICC) of 0.8.