[Quantization Methodology of Autofluorescence Bronchoscopy Image in the YUV System].

Xiaoxuan Zheng,Hongkai Xiong,Yong Li,Baohui Han,Jiayuan Sun
DOI: https://doi.org/10.3779/j.issn.1009-3419.2014.11.05
2014-01-01
Abstract:BACKGROUND AND OBJECTIVES:The aim of this study is to determine the best reference values of the optimal evaluation indexes that identify different disease types. Disease identification was conducted using the YUV quantitative analysis of autofluorescence bronchoscopy (AFB) images in the target areas. Furthermore, this study discusses the significance of AFB in the diagnosis of the central-type lung cancer.METHODS:A biopsy was conducted for cases that showed pathologic changes under either autofluorescence or white-light bronchoscopy. Moreover, MATLAB was used to carry out the quantitative analyses of lesion in multi-color spaces from AFB images. The cases were divided into different groups according to the pathological diagnosis of normal bronchial mucosa, inflammation, low-grade dysplasia (LGD), high-grade dysplasia (HGD), and invasive cancer. SPSS 11.5 was used to process the data for statistical analysis.RESULTS:The Y values were different and statistically different between invasive cancer and LGD (P<0.001) and invasive cancer and inflammation (P=0.040), respectively. The U values between invasive cancer and the other groups were statistically different (P<0.050). Similarly, the V values between invasive cancer and LGD and inflammation and normal bronchial mucosa were different. Lastly, the V values between normal bronchial mucosa and HGD and inflammation and normal bronchial mucosa were different.CONCLUSIONS:The YUV values in the AFB effectively identified benign and malignant diseases and were proven to be effective scientific bases for the accurate AFB diagnosis of lung cancer.
What problem does this paper attempt to address?