Detection and diagnosis of human oral cancer using hypericin fluorescence endoscopic imaging interfaced with embedded computing

Oral cancers are currently diagnosed using white light endoscopy and histopathology. However, oral tumours are mostly superficial and can be difficult to visualise. Here we present the use of hypericin with fluorescence endoscopy and laser confocal fluorescence endomicroscopy interfaced with embedded computing for the diagnosis of oral cancers. Fluorescence imaging of oral lesions was carried out in the clinic using a fluorescence endoscope. The images were analyzed to extract the red to blue (R/B) ratios to discriminate between tissue types. The results showed that the R/B ratio is a good image parameter to discriminate between normal, hyperplastic and malignant oral tissue. We are also developing an embedded, real-time computing system interfaced to a fluorescence endomicroscope for 3D visualization of tumors, where synchronization of cross-sectional image grabbing and Z-depth scanning is realized through programming a Field-Programmable Gate Array. In addition to the programming task, a proprietary control circuit has been developed for the automated 3D reconstruction of fluorescence sections; and preliminary results from fluorescent samples have demonstrated the potential of this system for real-time in vivo 3D visualization of tumours. This will ultimately enable same-day clinical diagnosis to be achieved and further enhance the clinical usefulness of fluorescence diagnostic imaging.