Infrared (IR) spectral microimaging is an efficient label-free optical method to analyze biological samples and to provide information about their intrinsic biochemical composition. Recent studies have shown its potential to detect and characterize cancerous tissues in their early stages, independently of visual morphology. IR microimaging appears more informative than conventional histology and could thus be developed as a sensitive, non-destructive and objective diagnostic tool in clinical oncology. The discrimination between tumoral and surrounding tissues relies on highlighting subtle IR spectral differences. An innovative fuzzy clustering based algorithm was thus developed in order to introduce the notion of nuance into the clustering of IR image pixels and to automatically determine the optimal clustering parameters, particularly the number of clusters. IR spectral imaging has been directly acquired on formalin-fixed paraffin-embedded tissue sections of human cutaneous melanomas with different degrees of invasiveness (n=10 patients). After the application of the fuzzy algorithm, the reconstructed color-coded clustering images allow recovering different histological structures automatically, particularly to precisely localize tumoral areas and their normal counterparts. However, more than reproducing classical histology, our algorithm can give access to additional information about the tissular structures. The notion of nuance introduced by fuzzy clustering is valuable for the pixels located at the interface between tumoral tissue and its microenvironment. Heterogeneous transitional areas between tumor and marginal normal tissue are identified, what is not possible on hematoxylin-eosin stained section or by conventional clustering. In addition, this automatic approach of IR data processing permits to highlight the interconnectivity between heterogeneous intratumoral areas and between tumoral and peritumoral structures. These preliminary results show significant potential for probing tumor progression and for early determination of tumor aggressiveness in cutaneous melanomas. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4052. doi:1538-7445.AM2012-4052