Bronchoscopic Fluorescence Image Enhancement Using Digital Image Processing Techniques

摘要

Based on the analysis of the characteristics of bronchoscopic fluorescence image, four kinds of digital image processing (DIP) techniques are presented to enhance the contrast of fluorescence image between early lung cancer and surrounding normal tissues. Firstly, nearest neighbor averaging method is used to smooth noise that comes from different sources, which mainly including the photoelectric and electronic noise in the diagnostic system. next, a method of enhancing contrast in fluorescence image is devised based on real-time digital subtraction of a background video image from a signal-plus-background video image. Then, segmental linearity image enhancement technique is applied after background subtraction to project the lesions. Finally, A processing technique for pseudocolor display of monochrome to enhance color contrast for the fluorescence image of early lung cancer, which highly enhanced the discrimination of lung cancer image in color. The validity of these approaches have been verified and primary clinical results show that a sufficient fluorescence contrast of suspicions versus normal tissue is obtained, which can effectively eliminate the false positive and negative diagnosis occurring in the clinical application.