The changes in the autofluorescence characteristics of thebronchial tissue is of crucial interest as a cancer diagnostic tool. Evidenceexists that this native fluorescence or autofluorescence of bronchialtissues changes when they turn dysplastic and to carcinoma insitu. There is good agreement that the lesions display a decrease ofautofluorescence in the green region of the spectrum under illuminationwith violet-light, and a relative increase in the red region of thespectrum is often reported. Imaging devices rely on this principle todetect early cancerous lesions in the bronchi. Based on a spectroscopicstudy, an industrial imaging prototype is developed to detectearly cancerous lesions in collaboration with the firm Richard WolfEndoskope GmbH, Germany. A preliminary clinical trial involving 20patients with this spectrally optimized system shows that the autofluorescencecan help to detect most lesions that would otherwise haveremained invisible to an experienced endoscopist under white lightillumination. A systematic off line analysis of the autofluorescenceimages pointed out that real-time decisional functions can be definedto reduce the number of false positive results. Using this method, apositive predictive value (PPV) of 75% is reached using autofluorescenceonly. Moreover, a PPV of 100% is obtained, when combiningthe white light (WL) mode and the autofluorescence (AF) mode, at theapplied conditions. Furthermore, the sensitivity is estimated to betwice higher in the AF mode than in WL mode.
|