Tags
No Tags
| Title: | A study on the use of polarized light in application to noninvasive tissue diagnostics |
| Author: | Li, Yanfang |
| Description: | In this dissertation, multiple investigations on the implementation and use of polarized light for noninvasive diagnostics in turbid media are reported on and discussed. The first study focuses on the development and construction of an automated Mueller matrix imaging system. In the first investigation, a method was developed to estimate the scattering coefficient contribution as a function of particle size in complex mixtures of polystyrene spheres. A partial least squares approach is used to accurately predict the individual scattering coefficient contributions in phantoms containing 0.2, 0.5, 1, and 2 µm diameter spheres. It is also shown how scattering type is encoded within the Mueller matrix. In addition, the backscattered Mueller matrix of cultured normal human melanocytes, HEMn-LP, and melanoma cells, WM115 and WM793, are investigated for potential cell characterization. These results indicate that the mitochondria contribute significantly to the backscattering polarization signals. In the second investigation, Mueller matrix imaging is applied for in vitro collagen degradation classification. Four different levels of collagen degradations were imposed on in vitro human dermis via collagenase type IA-s. Polar decomposition of the Mueller matrix was performed to better interpret the image-based measurements for the optical differentiation between the degraded and normal dermis. Combining these results in conjunction with the Mueller matrix elements, the extracted features were analyzed by classification tree to develop an accurate classification model. Such methods may eventually lead to the development of improved diagnostic tools capable of characterizing and distinguishing between tissue abnormalities. The final investigation involves the application of diffuse reflectance polarization imaging to detect glucose concentration levels in highly scattering turbid media. Through the application of partial least squares regression, it is shown that accurate glucose predication in highly scattering media can be achieved. Furthermore, through the use of principal component analysis, the correlation between changes in the scattering coefficient and optical activity in relation to glucose concentration within the polarization image measurements is demonstrated. The findings of this basic research could be potentially extended to biological tissue for the eventual development of a noninvasive physiological glucose monitor. |
| Permanent Link: |
http://rave.ohiolink.edu/etdc/view?acc_num=toledo1134596719
http://hdl.handle.net/2374.OX/103894 |
| Date: | 2005 |
| Files | Size | Format | View |
|---|---|---|---|
|
There are no files associated with this item. |
|||