패턴 조사 현미경을 이용한 3차원 내시경 연구

Abstract

An endoscope is essential for observing the inside of the human body such as stomach and intestine to investigate any kinds of diseases without surgery and has been also used as an assistant device for a surgeon to look into surgical area in detail. In addition to medical applications, an endoscope or a boroscope is widely used in industrial fields for measurements and inspections of small defects through small holes. Even though interferometry and microscopy have become representative technologies obtaining 3D surface profile of a target as well as 2D features of a specimen, their typical microscopic configurations are not appropriate when the target is located inside a certain object because of their relatively bulky probes. Therefore, endoscopy or boroscopy is a promising technology to observe a target through a tiny hole in combination of imaging optics in medical and industrial fields because a small size of distal end can go through a small hole and the optical imaging fiber bundle can transfer the image to the imaging device. Since a typical endoscope acquires 2D images only, however, it cannot provide the height and depth information in the acquired image. To overcome this limitation, several types of the endoscopes have been developed through introducing conventional 3D measurement technologies such as stereoscopy and optical triangulation. A stereoscopic endoscope observes the target with two different sight angles like human eyes and it can achieve 3D imaging by the combination of two 2D images. It has the advantage of a large field of view and relatively simple configuration, but the measurement resolution is not so high because it has been developed as an observation tool. An endoscope based on the optical triangulation can improve the measurement resolution by the distant detection path from the illumination path but it is not avoidable to increase the size of the probe. Moreover, the conventional 3D endoscopes have the limitation for measurements of adjacent targets from the distal probes. On the other hand, the probe of the typical endoscope consists of two fiber bundles for illumination and image detection except the additional motion part to change the position and direction of the probe. The illumination and detection pathways are completely distinguished and the size of the distal probe should increase up to a few tens of millimeters, which limits applications of the endoscope when the target should be measured through extremely small. In this thesis, I propose an endoscopic technique for precise 3D imaging for medical and industrial applications. As the fundamental principle of 3D measurements, we adopt the continuously scanning structured illumination microscopy (CSSIM), which enables to obtain 3D sectional images by axial scanning. In order to reduce the size of distal probe end, the illumination and detection paths are designed as coaxial and a coherent imaging fiber bundle is used for transferring the illumination pattern to the target and vice versa. The proposed 3D endoscope consists of two parts, CSSIM and probe parts. In the CSSIM part, the incoherent light preventing coherent noises such as diffractions and speckles, is illuminated on a sinusoidal amplitude grating or a grid and the structured light pattern is projected on the proximal end surface of the imaging fiber bundle. Then, the structured light is delivered by the flexible imaging fiber bundle and transferred to the sample surface by the imaging lens of the probe part. When the light is reflected off on the target, it returns back by the imaging fiber bundle and goes to an imaging device. In this case, the whole system has 5 conjugate image planes; the grid surface, the proximal and distal ends of the imaging fiber bundles, the target surface and the imaging device plane. When the target is located on the best focal position, the structured light pattern can be projected on the target surface clearly and can be seen in the imaging device as well. In the imaging device, the obtained image contains the light pattern and the image of the target surface. When the target is axially moved from the focal position, both of the light pattern and the target image are blurred. Therefore, 3D imaging can be achieved by the detection of the highest visibility of the light pattern at each pixel of the imaging device. In order to minimize the probe size, I designed the illumination and detection path coaxially and a GRIN lens was used as the probe imaging lens. In order to improve the contrast of the sinusoidal illumination pattern reflected off on the target, we used polarizing optical components and confirmed that the visibility of the pattern was able to be used in CSSIM. The gauge block specimen was used for the verification of the performance and it was confirmed that the 3D surface profile was successfully measured with 16.1 μm repeatability for gauge block specimen. As a measurement example, a Korean coin was measured and the 3D surface profile was also well reconstructed.