High spatial frequency structured light imaging texture analysis using Gabor filtering differentiates tumor from normal tissue subtypes

Samuel S. Streeter; Benjamin W. Maloney; David M. McClatchy; Elizabeth J. Rizzo; Wendy A. Wells; Keith D. Paulsen; Brian W. Pogue

doi:10.1117/12.2507247

1 March 2019 High spatial frequency structured light imaging texture analysis using Gabor filtering differentiates tumor from normal tissue subtypes

Samuel S. Streeter, Benjamin W. Maloney, David M. McClatchy, Elizabeth J. Rizzo, Wendy A. Wells, Keith D. Paulsen, Brian W. Pogue

Author Affiliations +

Proceedings Volume 10874, Optical Tomography and Spectroscopy of Tissue XIII; 108740I (2019) https://doi.org/10.1117/12.2507247
Event: SPIE BiOS, 2019, San Francisco, California, United States

Abstract

A recent clinical study involved multi-spatial frequency (0.00, 0.15, 0.61, 1.37 mm^-1 ), multi-spectral (490, 550, 600, 700 nm) structured light imaging (SLI) of freshly excised, bread-loafed breast conserving surgery tumor specimens. The specimens contained regions of interest (ROIs) of confirmed and homogeneous histological categories of breast tumor tissue determined by expert histopathological analysis. ROIs were sampled into ~4x4mm sub-images before conversion into Gabor filter bank feature vectors. Feature vectors were used in binary, benign-malignant classification scenarios using a support vector machine classifier and 8-fold cross validation. Classification was performed first on feature vectors containing only planar (0.00 mm^-1 ) 490 nm monochromatic SLI data. For comparison, a second set of feature vectors from the same sub-images contained the same planar 490 nm illumination data concatenated with high spatial frequency (1.37 mm^-1 ) 490 nm monochromatic SLI data. The classification performance of each filter bank was determined for both sets of feature vectors based on receiver operating characteristic (ROC) area under the curve (AUC) values. Gabor filtering and subsequent classification revealed that surface tissue textures exhibited strong rotational dependence and that high spatial frequency surface tissue features were more diagnostic than low spatial frequency surface tissue features. A Gabor filter bank that included 4 relatively high spatial frequencies (0.70-1.98 mm^-1 ) and 8 rotation angles (0-157.5°, equally spaced) achieved ROC AUC values <0.9 for 9 of 12 binary tissue subtype classifications.

Conference Presentation

Citation Download Citation

Samuel S. Streeter, Benjamin W. Maloney, David M. McClatchy, Elizabeth J. Rizzo, Wendy A. Wells, Keith D. Paulsen, and Brian W. Pogue "High spatial frequency structured light imaging texture analysis using Gabor filtering differentiates tumor from normal tissue subtypes", Proc. SPIE 10874, Optical Tomography and Spectroscopy of Tissue XIII, 108740I (1 March 2019); https://doi.org/10.1117/12.2507247

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available