Infrared chemical imaging for pathology and forensic biology

Publication Type:
Thesis
Issue Date:
2013
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail01front.pdf245.05 kB
Adobe PDF
Thumbnail02whole.pdf21.12 MB
Adobe PDF
The objective of this research was to explore the capability of Fourier Transform Infrared Chemical Imaging (FTIR CI) for two specific pathology applications: 1) the analysis of human tissues for the diagnosis of melanoma, and 2) incised skin wound age determination for the purpose of forensic investigation. For the melanoma study, thin serial sections were obtained from an archival tissue bank that consisted of pathologist pre-diagnosed (“gold-standard”), paraffin-embedded, human skin and lymph node tissues. Thin sections from each block were mounted on infrared reflective microscope slides and imaged, and a selection of the total images nominated as either training or test samples. Each training sample image was then compared to its corresponding haematoxylin and eosin (H&E)-stained section and reference library spectra extracted. Vertex component analysis (VCA) as a spectral feature extraction method was also explored. Classification of the test sample images was then performed using the spectral angle mapper (SAM) algorithm and the accuracy assessed by comparing the resulting classification images to the H&E-stained tissue sections. The tissue classification model developed produced a range in result quality, and highlighted various critical aspects in the construction of such methodologies. The taking of spectral derivatives improved image classification, as did the removal of paraffin from the tissue (although no data treatment targeting the paraffin was conducted on the non-deparaffinised tissues). Although the accuracy achieved in this study fell short of that required for clinical practice, the results obtained demonstrate that further investigation into the SAM algorithm as a tissue classifying tool is certainly warranted. The second pathology application explored the ageing of wounds, a determination that may be critical in criminal investigations, particularly in homicide investigations, in which the timing of wound infliction may be crucial evidence. For this study, incised wounds were inflicted on rats in a controlled manner and the tissue excised following a known amount of healing time ranging from 5 minutes to 288 hours (12 days). Thin sections of the wounds were mounted on infrared reflective slides, deparaffinised and then imaged using FTIR CI. Although four classification models were attempted, none were capable of producing highly accurate wound age determination. Spectral variation was observed between earlier and later wound ages using some of the classification methods, but the ability to correctly group the test samples into their respective age groups was not achieved. Based on the number of variables which must be taken into consideration when performing such a study, and the number of areas identified as needing further improvement (e.g. spectral data quality), the fact that even a limited form of discrimination was achieved using FTIR CI was encouraging.
Please use this identifier to cite or link to this item: