Texture analysis of optical coherence tomography images: feasibility for tissue classification

Area of Science:

• Biomedical Imaging
• Medical Diagnostics
• Image Analysis

Background:

• Optical coherence tomography (OCT) provides cross-sectional tissue images by analyzing back-reflected light.
• Standard OCT imaging resolution (10-15 µm) is optimal for structures tens to hundreds of microns in size.
• Many tissues lack discernible structures in this range, limiting OCT's apparent utility.

Purpose of the Study:

• To evaluate the efficacy of statistical and spectral texture analysis for differentiating tissue types using OCT images.
• To determine if OCT image texture, particularly speckle, can be used for tissue identification in the absence of visible structures.

Main Methods:

• Application of statistical and spectral texture analysis techniques to OCT images.
• Classification of different tissue types based on extracted textural features and speckle content.
• Evaluation of classification accuracy for visually distinct and similar tissue types.

Main Results:

• High correct classification rates for visually distinct tissues (e.g., mouse skin and fat at 98.5% and 97.3%).
• Reasonable classification rates for visually similar tissues (normal vs. abnormal mouse lung at 64.0% and 88.6%).
• Demonstration that texture analysis can differentiate tissues irrespective of visible structural content.

Conclusions:

• Texture analysis of OCT images, leveraging speckle patterns, offers a viable method for tissue differentiation.
• This approach expands the applicability of OCT to tissues lacking prominent visible structures.
• OCT texture analysis shows potential for non-reliance on visible structures in tissue identification.