Paper
28 February 2012 Multispectral image enhancement by spectral shifting
Author Affiliations +
Abstract
A multispectral enhancement method that preserves the natural color of the background pixels was previously proposed. In such method, the band for enhancement was identified from the N-band spectral residual-error of the objects of interest. The spectral residual-error is determined by taking the difference between the original spectrum of the pixel and its estimate using M<<N principal components in principal component analysis (PCA). However, for stained histopathology images where staining variations do exist even among tissue sections the band for enhancement could vary. In this work, we introduced a modification to the previously proposed multispectral enhancement method such that the band for enhancement could be specified independent of the spectral residualerror configurations. In the proposed modification the original spectral transmittance of the pixels at each band are shifted by the product between the spectral residual-error coefficient, which is the most dominant PC coefficient of the spectral error, of the pixel and the weighting factor assigned by the user to each band. Results of our experiments on H&E stained sections of liver tissue show that the proposed modification delivers more consistent enhancement results compared to the previously proposed methods, especially when the band for enhancement varies.
© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Pinky A. Bautista and Yukako Yagi "Multispectral image enhancement by spectral shifting", Proc. SPIE 8214, Advanced Biomedical and Clinical Diagnostic Systems X, 82140K (28 February 2012); https://doi.org/10.1117/12.910060
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Image enhancement

Transmittance

Tissues

Multispectral imaging

Collagen

Principal component analysis

Visualization

Back to Top