TY - JOUR
T1 - Auto-detection of cervical collagen and elastin in Mueller matrix polarimetry microscopic images using K-NN and semantic segmentation classification
AU - Roa, Camilo
AU - Du Le, V. N.
AU - Mahendroo, Mala
AU - Saytashev, Ilyas
AU - Ramella-Roman, Jessica C.
N1 - Funding Information:
National Science Foundation (DMR 1548924).
Funding Information:
Acknowledgments. This work was also supported by Herbert Wertheim College of Medicine and the Herbert and Nicole Wertheim Professorship Endowment. We are grateful to Dr. Timothy Allen for access to the Olympus microscope. We thank Dr. Mariano Colon-Caraballo at UT Southwestern Medical Center for guidance on elastin immunohistochemistry.
Publisher Copyright:
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement.
PY - 2021/4
Y1 - 2021/4
N2 - We propose an approach for discriminating fibrillar collagen fibers from elastic fibers in the mouse cervix in Mueller matrix microscopy using convolutional neural networks (CNN) and K-nearest neighbor (K-NN) for classification. Second harmonic generation (SHG), two-photon excitation fluorescence (TPEF), and Mueller matrix polarimetry images of the mice cervix were collected with a self-validating Mueller matrix micro-mesoscope (SAMMM) system. The components and decompositions of each Mueller matrix were arranged as individual channels of information, forming one 3-D voxel per cervical slice. The classification algorithms analyzed each voxel and determined the amount of collagen and elastin, pixel by pixel, on each slice. SHG and TPEF were used as ground truths. To assess the accuracy of the results, mean-square error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) were used. Although the training and testing is limited to 11 and 5 cervical slices, respectively, MSE accuracy was above 85%, SNR was greater than 40 dB, and SSIM was larger than 90%.
AB - We propose an approach for discriminating fibrillar collagen fibers from elastic fibers in the mouse cervix in Mueller matrix microscopy using convolutional neural networks (CNN) and K-nearest neighbor (K-NN) for classification. Second harmonic generation (SHG), two-photon excitation fluorescence (TPEF), and Mueller matrix polarimetry images of the mice cervix were collected with a self-validating Mueller matrix micro-mesoscope (SAMMM) system. The components and decompositions of each Mueller matrix were arranged as individual channels of information, forming one 3-D voxel per cervical slice. The classification algorithms analyzed each voxel and determined the amount of collagen and elastin, pixel by pixel, on each slice. SHG and TPEF were used as ground truths. To assess the accuracy of the results, mean-square error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) were used. Although the training and testing is limited to 11 and 5 cervical slices, respectively, MSE accuracy was above 85%, SNR was greater than 40 dB, and SSIM was larger than 90%.
UR - http://www.scopus.com/inward/record.url?scp=85104442490&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85104442490&partnerID=8YFLogxK
U2 - 10.1364/BOE.420079
DO - 10.1364/BOE.420079
M3 - Article
C2 - 33996226
AN - SCOPUS:85104442490
SN - 2156-7085
VL - 12
SP - 2236
EP - 2249
JO - Biomedical Optics Express
JF - Biomedical Optics Express
IS - 4
ER -