Deifnition Based Cell Microscopic Image Segmentation and Counting Algorithm / 中国医学影像学杂志

ABSTRACT

Purpose To propose a definition based algorithm for segmenting and counting cell microscopic images. Materials and Methods Cell microscopic images were first pretreated and then transformed using discrete cosine transformation (DCT). The high frequency part was truncated and re-converted to differentiate clear and blurred images. The clear foreground regions were obtained. The intact objective was extracted using region growing method. Statistics and analysis of cell number were then conducted. Results This algorithm showed good performance in cell microscopic image counting with accuracy of over 90% at less than 100 ms/image. Conclusion Definition based method is fast and accurate in cell microscopic image segmentation and counting.