Microscopic Image Processing System Implementation And Algorithm Research

With the development of technology in image processing, network and database, digital microscopic medical images are now frequently used in medical areas. The development of microscopic medical image processing systems is a very important section in the way of medical modernization.This paper focuses on the development of microscopic image processing system and cell image segmentation algorithm. To separate cells - especially overlapping cells - is important but difficult. In order to effectively and precisely extract cells, several novel algorithms are presented in this paper. To produce a modified nonlinear distance image, an iterative erosion method with a dynamic structure rather than a fixed one is used in the algorithm. A segmentation method based on such dynamical iterative erosion method is then proposed. Based on the improved erosion method, a novel B-spline active contour model, whose energy terms depend not only on the cell image itself, but also on a modified nonlinear distance image, is also proposed to precisely separate cells. Experimental results show that the algorithm is effective for cell image segmentation.Chapter 1 gives the introduction of basic concepts in medical image processing and microscopic medical image analysis systems.Chapter 2 is about the design of our microscopic medical image analysis system. In this chapter, the hardware and software constructions and execution flow are presented. This chapter also introduces the image techniques that frequently used in microscopic medical image processing. Finally, this chapter gives an example of designing a subsystem module of the microscopic medical image analysis system.Chapter 3 firstly points out a shortcoming of classical iterative erosion. To improve the iterative erosion method, a new erosion algorithm which uses a dynamical iterative structure is then proposed. A morphological segmentation method, with the concept of largest separate area, is also proposed in this chapter.Chapter 4 points out the difficulties encountered by classical snake model and classical B-snake model in microscopic cell images. This chapter then presents the morphologically restricted B-snake model for cell image segmentation in detail.Chapter 5 concludes the whole paper.