Infrared imaging to predict oral cancer development in oral epithelial dysplasia


Year:

Session type:

Theme:

Barnaby Ellis1,Conor Whitley1,James Ingham1,Steve Barrett1,Peter Gardner2,Richard Shaw1,Peter Weightman1,Janet Risk1
1University of Liverpool,2University of Manchester

Abstract

Background

Oral squamous cell carcinoma (OSCC) is the sixth most common malignancy worldwide. Oral epithelial dysplasia (OED) is the most common precursor to OSCC, but a key clinical challenge is the identification of OED lesions with the capacity to transform to OSCC, as this is not defined with certainty by clinical or pathological methods. Our previous work has demonstrated the capacity of Fourier transform infrared (FTIR) imaging to distinguish OSCC from surrounding non-malignant tissue. We now demonstrate that this tool is also capable of predicting transformation in OED.

Method

Five micrometre sections from 5 transforming and 6 non-transforming, moderately dysplastic, OED lesions were cut onto calcium fluoride discs and imaged by FTIR. Samples were obtained from archival blocks less than 12 months prior to transformation (transforming lesions) or from lesions with at least 2 years follow up with no transformation (non-transforming lesions). Analysis used multivariate metric analysis1, PCA-LDA and other methods with the intention of building an interpretable, supervised model which can automatically discriminate between the two populations.

Results

Data was collected at >1500 IR wavelengths at a spatial resolution of ~5┬Ám. By using 4/5 of the data as a training set and 1/5 as a validation set (in all combinations), discrimination between transforming and non-transforming lesions in this single, WHO defined, class of OED was achieved at a sensitivity and specificity of >75% by all methods, demonstrating the utility of this technique. Key wavelengths for discrimination were identified.

Conclusion

FTIR discrimination between transforming and non-transforming moderately dysplastic OED out-performed oral pathological assessment in this small sample set and show promise as a method for early diagnosis and treatment of OSCC.

 1: J Ingham et al Infrared Physics & Technology 2019