Early detection of human head and neck cancer using in vivo multimodality microscopic imaging: Initial validation in a hamster model

Objectives To evaluate the performance of a multimodality probe - capable of performing label-free, ionizing radiation-free, co-registered, optical time-resolved fluorescence spectroscopy (TRFS), ultrasonic backscatter microscopy (UBM) and photo-acoustic imaging (PAI) - for early detection and staging of head and neck squamous cell carcinoma.

Methods An optical fiber-based TRFS system was integrated with a single element 41 MHz UBM system (spatial resolution <0.06 mm) by our group to construct the multimodality probe. The right cheek of 14 Syrian hamsters was treated with DMBA to induce carcinogenesis while 6 hamsters acted as healthy controls. For imaging (conducted in the 14th week after treatment), the cheek pouch of the anesthetized animals was everted to expose the region of interest (ROI) - as diagnosed by an otolaryngologist - and scanned by the probe in a raster fashion. Scanning was also conducted of the contra-lateral cheek and of control subjects. A biopsy from the imaged ROI was used for histopathological analysis.

Results With histopathology as the ground truth, TRFS data (depth sensitivity <0.3 mm) revealed high sensitivity to changes in the composition of intrinsic tissue fluorophores collagen, elastin, NADH, and porphyrins in early lesions and advanced cancer as opposed to healthy tissue. UBM images (depth sensitivity ~5-6 mm) provided a co-registered anatomical picture revealing epithelial, mucosal and sub-mucosal hypertrophy (early stage), and neoplastic changes such as lesion development and destruction of the structural integrity of tissue layers (advanced stage). Photoacoustic images acquired for the same ROI showed optical absorption contrast due to neovascularization.

Conclusions These initial studies confirmed the ability of the multimodality probe for differentiating normal mucosa from hyperplastic and cancerous tissues in the animal model. The investigated technology has potential to be directly translated to the human clinic.

Research Support This work was funded by the NIH grants R21-RR0025818, R01-HL67377, and the 2011 UC Davis Health System Vision Grant