To improve the accuracy and efficiency of medical diagnosis, an image analysis theory based on morphology of microscopic cells is proposed to complete image classification, identification and analysis. Firstly, the image edge detection algorithm and watershed segmentation algorithm are introduced in this paper. An integral image processing method of microscopic cell based on morphology is designed, which is an effective solution to the problem of uneven illumination and stained spots and other issues arising in the image processing. Then, in the image analysis stage, the morphology image analysis of microscopic cell is applied to diagnosis the hematological disease, simultaneously, the number of cells is calculated, morphological parameters are extracted and verification results are presented. Finally, a preliminary theoretical attempt about the medical diagnostic cell disease is made, and compared with actual value, the research results error is less than 3%. Experiments show that the proposed image analysis theory has a certain value in medical diagnosis on cell disease.