Detection of neoplasms in the oral cavity by digitized endoscopic imaging of 5-aminolevulinic acid-induced protoporphyrin IX fluorescence

Abstract

Studies of 5-aminolevulinic acid (5-ALA) induced protoporphyrin IX (PPIX) fluorescence have shown a sensitivity of 95-100% for oral cancer diagnosis, but the specificity is only about 50-60%. To improve the diagnostic specificity, a 5-ALA mediated digitized fluorescence endoscopic imaging system was built in this study to enable the on-line image acquisition, analysis and fluorescence quantification for the early detection of neoplasms in the oral cavity. PPIX fluorescence endoscopy and fluorescence image quantification were performed on 16 patients with known or suspected premalignant or malignant lesions in the oral cavity. Preliminary data from the head and neck clinical trials show that the red-to-blue intensity ratio of malignant tissue is larger than that of benign tissue. By applying the intensity ratio as a diagnostic algorithm, both high specificity and sensitivity can be achieved for discriminating malignant tissue from benign tissue, and the capability of classifying different histopathological stages of oral lesions has also been demonstrated. It was found that the red fluorescence intensity distribution in the lesion area could also be obtained to better understand the situation of PPIX accumulations in the tissues. Our initial results indicate that the digitized endoscopic imaging system combined with the fluorescence image quantification method and the ratio diagnostic algorithm developed in this study has the potential to significantly improve the non-invasive diagnosis of early oral neoplasms in vivo.