21 February 2018 Application of wavelet based MFDFA on Mueller matrix images for cervical pre-cancer detection
Author Affiliations +
A systematic study has been conducted on application of wavelet based multifractal de-trended fluctuation analysis (MFDFA) on Mueller matrix (MM) images of cervical tissue sections for early cancer detection. Changes in multiple scattering and orientation of fibers are observed by utilizing a discrete wavelet transform (Daubechies) which identifies fluctuations over polynomial trends. Fluctuation profiles, after 9th level decomposition, for all elements of MM qualitatively establish a demarcation of different grades of cancer from normal tissue. Moreover, applying MFDFA on MM images, Hurst exponent profiles for images of MM qualitatively are seen to display differences. In addition, the values of Hurst exponent increase for the diagonal elements of MM with increasing grades of the cervical cancer, while the value for the elements which correspond to linear polarizance decrease. However, for circular polarizance the value increases with increasing grades. These fluctuation profiles reveal the trend of local variation of refractive -indices and along with Hurst exponent profile, may serve as a useful biological metric in the early detection of cervical cancer. The quantitative measurements of Hurst exponent for diagonal and first column (polarizance governing elements) elements which reflect changes in multiple scattering and structural anisotropy in stroma, may be sensitive indicators of pre-cancer.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mohammad Zaffar, Mohammad Zaffar, Asima Pradhan, Asima Pradhan, } "Application of wavelet based MFDFA on Mueller matrix images for cervical pre-cancer detection", Proc. SPIE 10489, Optical Biopsy XVI: Toward Real-Time Spectroscopic Imaging and Diagnosis, 1048916 (21 February 2018); doi: 10.1117/12.2289965; https://doi.org/10.1117/12.2289965

Back to Top