Optimization of the spectral design used to detect early carcinoma in the human tracheo-bronchial tree by autofluorescence imaging

The early detection and localization of bronchial cancer remains a challenging task. Autofluorescence bronchoscopy is emerging as a useful diagnostic tool with improved sensitivity and specificity. Evidence exists that the native fluorescence or autofluorescence of bronchial tissues changes when they turn dysplastic or to carcinoma in situ (CIS). Early lesions in the bronchi tend to show a decrease in autofluorescence in the green region of the spectrum when excited with violet light and a relative increase in the red region of the spectrum. Several endoscopic imaging devices relying on these optical properties of bronchial mucosa have been developed. An industrial endoscopic autofluorescence imaging system for the detection of early cancerous lesions in the bronchi has been developed in collaboration with the firm Richard Wolf Endoskope GmbH, Knittlingen (Germany) and its performance has been evaluated in a previous clinical study. A second study, presented in this article, aims to optimize the spectral design of the device. Twenty-four lung cancer or high risk patients were enrolled in this study to assess the influence of additional backscattered red light on the tumor-to-healthy tissue contrast and to compare the effect of a narrow band violet excitation to a large band violet excitation. In our study we observed a three times higher contrast between cancer and healthy tissue, when backscattered red light was added to the violet excitation. The comparison between a narrow and a large band violet excitation indicated an increase of the tumor-to-healthy tissue contrast by the narrow band excitation.