Infrared imaging of oral squamous cell carcinoma can distinguish carcinoma from other cell types
Session type: Proffered paper sessions
Theme: Diagnosis and therapy
Fourier transformed infrared (FTIR) imaging is capable of grading prostate and ovarian cancer and can yield information about aggressiveness/staging. The oral cavity has a recognised pre-neoplastic condition and clinically unresolved staging conundrums. We hypothesise that infrared spectroscopy will be able to distinguish cancer from non-cancer cells in the oral cavity to augment clinical/pathological staging.
As proof of principle, 5µm sections from 7 oral cancer tissues containing both carcinoma and adjacent normal tissue were mounted onto calcium fluoride discs and 4.46mm2 areas of both tissue types were scanned at 1505 IR wavelengths at a resolution of 5µm. Areas containing carcinoma, normal and lymphoid cells were highlighted on the composite IR images with reference to H&E stained sections and the data was used to train and test a novel spectral algorithm to discriminate 7 different cell types: cancerous epithelial, normal epithelial, tumour stroma, normal stroma, metastatic tumour in lymph node, normal lymphoid, skeletal muscle.
Confidence values for cell type recognition was between 75 and 98% using 10-50 wavelength metrics, with cancerous epithelial cells being discriminated at a success rate of 86% using 50 metrics. The trained algorithm was then able to identify cell types in unlabelled images.
Carcinoma tissue can be identified with a confidence of >85% in both primary and lymph node metastatic sites using an algorithm trained by only a small sample set. This methodology has potential as an adjunct to pathological staging for identification of oral squamous cell carcinoma in the setting of: monitoring oral dysplasia transformation; surgical bed involvement; lymph node micrometastasis; identification of OSCC for neo-adjuvant treatment