The Key Technologies, Design And Implementation Of Image Analysis Platform For Lung Cancer Diagnosis Based On Near Infrared

Near infrared imaging technology aided the pathological organization of positioning is the advanced technology of computer aided diagnosis of one of the research focus and direction. It is an integrated optics, computer hardware and software technology, histopathology, multidisciplinary cross research direction of artificial intelligence and so on. Using molecular imaging agent indocyanine green fluorescence under 830 nm near-infrared light irradiation, and through the sensor imaging, via the backend computer software platform for analysis and image processing techniques to obtain pathological target parts of the image segmentation, according to the results and help for clinical medicine, surgery and medical research.This dissertation designed the diagnosis for lung cancer near infrared image analysis platform, combined with surgery navigator hardware, designed to assist doctors to position the sentinel lymph node in the process of operation. Near infrared image analysis platform is a software of image processing and display system, he can segment will target parts of the patients and displayed. Overall use of MVC design pattern, the platform adopts the standard of three-tier architecture, at the same time combined with Qt interface design framework, OpenCV open-source platform and advanced technology such as medical image processing method, integrates the advantage of laboratory related algorithm, can achieve better practicability. Function modules of the platform include image fusion module, image analysis module, image display module and image storage module four parts, respectively for image preprocessing, fusion, segmentation, display and storage. In addition, in this paper, two technical difficulties in the platform and research, the first is the near infrared near infrared image and visible light image fusion multimodal image registration, this dissertation designed via DOG filter with image sub-block based Surf feature image registration algorithm, better solved the problem of image fusion under different modal. The second is the near-infrared light under fluorescent area segmentation, image segmentation based on level set method is used to segment the unknown and highlighting the characteristics of the area, to the doctor to clinical guidance. Compared with the threshold segmentation method, the experiments show that it can more accurately target area for the position and contour is given.Due to the use of near infrared imaging in lung cancer sentinel lymph node localization, only small sample experiment abroad, domestic is completely blank. The system did some research in this aspect. At the same time the system is used in some of the improvement of medical image processing algorithm, to solve some practical problems in image segmentation.