Assessment of discriminant models in infrared imaging using constrained repeated random sampling – Cross validation

• Published in: Analytica chimica acta 2018-11, Vol.1033, p.156-164
• Main Authors: Pérez-Guaita, David, Kuligowski, Julia, Lendl, Bernhard, Wood, Bayden R., Quintás, Guillermo
• Format: Article
• Language: English
• Subjects: Constrained repeated random sampling - cross validation; Cross validation; Discriminant Analysis; Infrared hyperspectral imaging; Infrared imaging; Infrared radiation; Infrared Rays; Microscopy; Nondestructive testing; Oversampling; Partial least squares-discriminant analysis; Pixels; Random sampling; Sampling; Software; Spatial data; Spatial resolution; Spectrophotometry, Infrared; Statistical sampling; Tissues; Training
• ISSN: 0003-2670; 1873-4324
• DOI: 10.1016/j.aca.2018.05.019

Infrared (IR) imaging is an emerging and powerful approach for studying the molecular composition of cells and tissues. It is a non-destructive and phenotypic technique which combines label-free molecular specific information from cells and tissues provided by IR with spatial resolution, offering great potential in biochemical and biomedical research and routine applications. The application of multivariate discriminant analysis using bilinear models such as Partial Least Squares-Discriminant Analysis (PLS-DA) to IR images requires to unfold the spatial directions in a two-way matrix, resulting in a loss of spatial information and structure. In this article, first we evidence that internal validation methods such as repeated k-fold cross-validation (CV) can be overly optimistic when the pixel size of the image is lower than the lateral spatial resolution. Secondly, we propose a new approach for the unbiased internal evaluation of the model performance named COnstrained Repeated Random Subsampling–Cross Validation (CORRS-CV). This method is based on the generation of q training and test sub-sets using a constrained random sampling of n training pixels without replacement and it circumvents overly optimistic effects due to oversampling, providing more accurate and robust images. The approach can be applied in IR microscopy for the development of discriminant models to analyse underlying biochemical differences associated to anatomical and histopathological features in cells and tissues. [Display omitted] •Validation of discriminant models in IR hyperspectral imaging is challenging.•Internal validation by e.g. repeated k-fold CV can be overly optimistic.•COnstrained Repeated Random Subsampling–Cross Validation (CORRS-CV) is proposed.•CORRS-CV uses a spatial constraint in the selection of the training and test sets.•CORRS-CV led to better, more accurate, estimates of model performance.