ELLIPSE - BASED TEMPLATE MATCHING FOR CELL IDENTIFICATION IN FLUORESCENCE MICROSCOPE IMAGES

Automatic Cell detection in fluorescence microscopy(FM)images is essential
operation in automated cell analysis for diagnosis and treatment of oral
cancer which includes the study of cell shapes.The biologists are able to
detect, count all the cells automatically in FM images which provide high reliability
and fasten the clinicians work. In this work, a cell detection method in FM images
obtained from Cancer Research Malaysia is proposed. This method is divided into two
main stages, cell enhancement, and cell detection. The aim of cell enhancement is to
remove unwanted noise and enhance cells boundary prior to cell detection operation.
This work presents a non-exhaustive comparison between several linear and non-linear
filters applied to FM images. The performance of the enhancement techniques is
evaluated and tested in terms of PSNR and ASNRon 30 FM images that exhibiting low
contrast and non-uniform illumination. Results from this experiment indicate that non-linear filters give better contrast and background removal than the linear filters.