Infrared Spectral Microimaging

Infrared (IR) spectral microimaging is a label-free optical technique that permits to probe the intrinsic chemical composition of biological samples. This technique appears more informative than conventional histology and could be developed as a non-destructive and objective diagnostic tool for in routine use in pathology departments. The discrimination between tumoral and neighbouring tissues relies on highlighting subtle spectral differences by means of advanced statistical multivariate data processing. In this way, we applied an innovative algorithm based on fuzzy clustering to take into account the notion of nuance into the clustering of IR image pixels. IR spectral imaging of human cutaneous melanomas was performed on paraffin-embedded tissue sections, without previous chemical dewaxing. Reconstructed colour-coded images allow recovering automatically different histological structures, by distinguishing tumour from the connective tissue. This approach gives also access to information about the tissue structures. Indeed, it permits to highlight the interconnection between neighbouring structures. In this work in progress study, the analysis was focussed on the interconnection between the tumoral and peritumoral structures and between the tumoral areas in case of a heterogeneous lesion. These first encouraging results pave the way to access new diagnostic and prognostic criteria associated to the tumour aggressiveness in cutaneous melanomas.