Article Text

Download PDFPDF
PTH-328 High resolution spectral imaging: novel diagnostics in colorectal cancer
  1. R Griggs1,
  2. J Nallala2,
  3. G Lloyd1,
  4. T Cook3,
  5. C Kendall1,
  6. H Barr1,3,
  7. N Shepherd1,4,
  8. N Stone2
  1. 1Biophotonics Research Unit, Gloucestershire Hospitals NHS Foundation Trust, Gloucester
  2. 2Physics Department, University of Exeter, Exeter
  3. 3Department of Surgery, Gloucestershire Hospitals NHS Foundation Trust, Gloucester
  4. 4Cellular Pathology, Gloucestershire Hospitals NHS Foundation Trust, Cheltenham, UK


Introduction Colorectal cancer is the second most common cause of cancer death in the UK and accounts for 13% of all new cancer cases each year.1Most of these arise from pre-existing adenomatous polyps.2Recent research has supported an early diagnosis and prevention approach with the aim of improving survival and reducing overall mortality from colorectal cancer death.3The Bowel Cancer Screening Programme (BCSP) was commenced to drive early diagnosis of colorectal cancer but generates a huge volume of tissue for analysis from biopsy and polypectomy. The aim of this study was to investigate and demonstrate the use of novel high resolution infrared spectroscopy in colonic tissue to differentiate and classify a representative range of abnormal pathologies.

Method 65 cases from 5 different pathology groups were identified from 2010 to 2014. These formalin-fixed and paraffin-embedded samples were routinely processed and thin sections placed on calcium fluoride slides. High resolution infrared imaging with a pixel resolution of 1.1 microns was performed on the tissue and the absorption spectra measured for each case. Accepted methods of multivariate spectral analysis were undertaken to elucidate the spectral differences across the dataset and classify the tissue spectra representative of the tissue biochemistry.

Results High resolution infrared spectral imaging was successfully performed on thin sections of unstained colonic tissue. Specimen analysis generated over 1.5 million spectra which were used to discriminate the 5 pathological groups. The epithelial spectra from each case were identified and then principal component and linear discriminant analysis applied, and a classification model built to demonstrate the spectral differences. Leave one sample out cross validation (LOSOCV) resulted in a sensitivity of 65% and specificity of 82.3% for the technique to successfully discriminate normal from cancerous tissue. Dysplasia and cancer were also successfully discriminated with a sensitivity of 75.3% and specificity of 61.3%.

Conclusion High resolution infrared spectral imaging is a novel technique which has been proven to be successful, reliable and rapid in the analysis of paraffin-embedded colonic pathology. It demonstrates suitability for automated analysis of specimens and therefore the potential for use as an adjunct in the diagnostic pathway for both early polyp and invasive cancer, particularly in high volume BCSP cases or where the traditional pathological diagnosis is time-consuming or difficult.

Disclosure of interest None Declared.



  2. Scholefield JH. BMJ. 2000;321:1004–6

  3. Winawer SJ, Fletcher RH, Miller L, et al. Gastroenterology. 1997;112:594–642

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.