Image segmentation using PCNN and template matching for blood cell counting

Biomedical image processing has experienced intense growth with interdisciplinary research field attracting expertise from computer science. Medical image processing assists doctors to diagnose diseases and minimizes the workload. Especially image segmentation is a technique that is useful to explore the details of image. Counting and segmenting red blood cells images are challenging task in the cellular embryology and pathology. Different neural network algorithms are used for image segmentation and counting such as Pulse Coupled Neural Network (PCNN), Multilayer Perceptron (MLP) neural networks etc. Applying these algorithms for segmenting and counting on overlapped blood cells images gives less accurate results and diagnosis is not very efficient. This paper proposes combination of PCNN and template matching algorithm to segment and count overlapped red blood cells image. The area of isolated red blood cell is considered as a template. Proposed algorithm performs better on overlapped red blood cells images in comparison with existing system, based on the experimental results.