Computer-aided diagnosis of ophthalmic diseases using optical coherence tomography (OCT) relies on the extraction
of thickness and size measures from the OCT images, but such defined layers are usually not observed
in emerging OCT applications aimed at "optical biopsy" such as pulmonology or gastroenterology. Mathematical
methods such as Principal Component Analysis (PCA) or textural analyses including both spatial textural
analysis derived from the two-dimensional discrete Fourier transform (DFT) and statistical texture analysis obtained
independently from center-symmetric auto-correlation (CSAC) and spatial grey-level dependency matrices
(SGLDM), as well as, quantitative measurements of the attenuation coefficient have been previously proposed
to overcome this problem. We recently proposed an alternative approach consisting of a region segmentation
according to the intensity variation along the vertical axis and a pure statistical technology for feature quantification.
OCT images were first segmented in the axial direction in an automated manner according to intensity.
Afterwards, a morphological analysis of the segmented OCT images was employed for quantifying the features
that served for tissue classification. In this study, a PCA processing of the extracted features is accomplished
to combine their discriminative power in a lower number of dimensions. Ready discrimination of gastrointestinal
surgical specimens is attained demonstrating that the approach further surpasses the algorithms previously
reported and is feasible for tissue classification in the clinical setting.