Skin cancer detection using laser imaging

Abstract

Endeavors to develop a non-invasive and reliable device to detect different types of skin cancer in its early stages have been undertaken over the past few years. Although some advances have been made towards such developments, there remains great scope for the development of a non-expensive, compact, reliable, safe, easy to use, fast, sensitive, and non-invasive device. This project will apply the innovative laser feedback interferometry (LFI) technique to skin cancer detection via laser imaging using near-infrared vertical cavity surface emitting lasers (VCSELs). In this project, the science behind light--tissue interactions has been explored and skin imaging modalities have been elaborately studied. As a result, LFI has been exploited in various configurations to image different modalities such as Doppler flowmetry and confocal reflectance. In addition to single modality imaging, dual-modality imaging has also been proposed and conducted experimentally, to image Doppler flowmetry and confocal reflectance at the same time, using an integrated LFI system. It is shown that dual-modality imaging can yield images with higher contrast which provide more information about morphological and functional characteristics of a target under test. In addition to two-dimensional LFI imaging, three-dimensional imaging systems have also been studied and employed in the laboratory. A tomographic image identifies tissue volume with altered optical properties due to cancerous or other malignant tissue processes. We applied the three-dimensional imaging system to different types of keratinocyte skin cancer phantoms and results show high level of sensitivity in providing tomographic images. Using high numerical aperture lenses, making images at resolutions down to micrometer scales is possible, which is promising for microscopy of biological structures. Furthermore, parallel LFI imaging using an array of 24 VCSELs was also studied theoretically and experimentally as a part of this work. A VCSEL array was used in a parallel read-out LFI imaging system to develop a fast and high resolution device, which eliminates the need to scan mechanically in one dimension. Investigating the complex interactions of the laser beam and tissue requires a powerful and robust simulation technique. Monte Carlo was used as a simulation technique to model the light--tissue interactions occurring in the optical systems in this work. Firstly, we tested and validated the Monte Carlo main engine which was developed in the group, by studying morphological aspects of the simulated and experimental signals in a laser Doppler velocimetry system, and deep insight into the nature of the LFI Doppler spectrum and further conclusions were drawn. Secondly, the Monte Carlo engine was used in modeling and analyzing of optical systems, in this work. We developed tissue phantom making techniques which were used to examine the experimental ideas in place of biological samples which are difficult to obtain and work with. Therefore, tissue phantom techniques were studied extensively and realistic models were made. We used agar gel and silicone based tissue phantoms to make multi-layer structures resembling skin tissue. We included deformities in these structures representative of cancerous tissue and controlled the optical properties of the agar samples by adding different doses of titanium dioxide and Indian ink as the main scatterer and absorber, respectively. Development in different areas of this research paves the way for building a non-invasive, safe, low-cost, compact, and sensitive laser imaging device, which may play an important role in early skin cancer detection. We believe results of this work indicate that LFI can be used for this purpose and the knowledge acquired during the course of this work will help us to apply this system to real biological tissues.