Hyperspectral (HS) imaging combines spectral measurement of a pixel with 2D imaging technology. It is capable to provide a series of images containing both spectral and spatial information, and has been widely used in medical domain. However, most researches on medical HS imaging are regarding ex-vivo biopsy or skin and oral mucosa. The study on HS imaging regarding in-vivo disease lags far behind.
In this thesis, we developed a novel flexible HS endoscope system. It is capable to obtain a series of HS images in vivo in a non-contact way among the wavelength range of 405 – 665 nm. After a lot of time-consuming modifying and debugging work, this new system has high stability and convenience to be applied in clinic now. We evaluated this system in clinic. First, we got ethics approval for clinical trials. Then, we obtained HS images regarding gastrointestinal (GI) diseases inside patients using this system. As far as we know, this type of in-vivo image data has not been reported in previous literatures. Thus using these HS images, we built a database for GI mucosa. Next, we analyzed some typical HS images tentatively. The method of Recursive Divergence is implemented to extract valuable and diagnostic information from these HS images. The results prove the effect and applicability of this new HS endoscope system, which has shown the great potential to be used as a platform and guidance for further medical studies. To further apply the analysis results in clinic, we propose a novel Adaptive Narrow-Band Imaging (ANBI) method based on band selection of HS images of a specific type of disease. It is expected that the new technique has higher accuracy, sensitivity, and specificity compared to conventional Narrow-Band Imaging (NBI) technique. In this thesis, we also discuss the future direction of the system improvement. Especially, to improve light intensity and signal-noise-ratio of HS images in wide-field view, we propose a new imaging method using broad- and overlapped-band filters. Although this method only performs greatly on the foundation of accurate image registration, we hope to apply it in our system in the future.
Identifer | oai:union.ndltd.org:GATECH/oai:smartech.gatech.edu:1853/55026 |
Date | 27 May 2016 |
Creators | Han, Zhimin |
Contributors | Dong, Shuxiang |
Publisher | Georgia Institute of Technology |
Source Sets | Georgia Tech Electronic Thesis and Dissertation Archive |
Language | en_US |
Detected Language | English |
Type | Dissertation |
Format | application/pdf |
Page generated in 0.001 seconds