Affiliations: Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, P.R. China | Pathology Laboratory, Institute of Tuberculosis and Thoracic Tumors, Beijing 101100, P.R. China | Chinese National Software and Technology Company, Beijing 100081, P.R. China
Note: [] Corresponding author: Kuang-Hu Hu, Institute of Biophysics Chinese Academy of Sciences, 15 Datun Road, Chaoyang District, Beijing 100101, P.R. China. Tel.: +86 10 64888589; Fax: +86 10 64871293; E-mail: hukh@sun5.ibp.ac.cn.
Abstract: Recognition of lung cancer cells is very important to the clinical diagnosis of lung cancer. In this paper we present a novel method to extract the structure characteristics of lung cancer cells and automatically recognize their types. Firstly soft mathematical morphology methods are used to enhance the grayscale image, to improve the definition of images, and to eliminate most of disturbance, noise and information of subordinate images, so the contour of target lung cancer cell and biological shape characteristic parameters can be extracted accurately. Then the minimum distance classifier is introduced to realize the automatic recognition of different types of lung cancer cells. A software system named “CANCER.LUNG” is established to demonstrate the efficiency of this method. The clinical experiments show that this method can accurately and objectively recognize the type of lung cancer cells, which can significantly improve the pathology research on the pathological changes of lung cancer and clinical assistant diagnoses.
Keywords: Lung cancer cells, mathematical morphology, classification and recognition, clinical assistant diagnoses
