| dc.description.abstract | Optical Coherence Tomography (OCT) has been used extensively for many fundamental research and clinical applications since it provides cross-sectional images with high resolution noninvasively. Moreover, phase-sensitive OCT (PhOCT) technique makes OCT more appealing by producing functional information such as blood flow, elastic properties, and vibrational information. Among them, we are interested in measuring vibration to investigate functions of the middle and inner ear. OCT based vibrometry has been explored using a swept-laser source because of its advantages in acquisition rate and imaging depth compared to an OCT system using a spectrometer. However, a swept-laser source carries inter- and intra-sweep variability which is a critical problems that negatively affects image quality and displacement sensitivity. Also, OCT vibrometry is vulnerable to a significant increase in processing time because of the considerable amount of data as well as longer processing steps that are required to obtain structural and vibrational information. Longer processing time can make OCT vibrometry less appealing to biologists and clinicians even though it provides useful information because of greatly reduced experimental and diagnostic throughputs. Finally, a theoretical framework has not been established to analyze the effects of additive noise or an adjacent reflector, either of which is intrinsic in OCT vibrometry, on vibratory measurements. Without this framework it is hard to evaluate the performance of a OCT vibrometry system and figure out methods to alleviate those negative effects. Therefore, we proposed methods in this work to solve these important issues that impede the progress in the use of swept-laser OCT based vibrometry. The first problem, intrinsic in a swept-laser source, was tackled by a signal processing approach that calibrated every sweep of the laser with a complex FIR filter and interpolation. In this approach, a complex FIR filter was adopted to extract the non-linear wavenumber from a reference signal in an effective way, and interpolation was employed to calibrate OCT signals to have the same linear wavenumber for removing intra- and inter-sweep variability. For real time processing, the proposed method was implemented on FPGA. This approach was compared to the traditional IFFT-FFT based spectral calibration and showed the use of less resources in FPGA and as good or slightly better OCT image quality and displacement sensitivity. Also, it was explored to compensate chromatic variation from a laser source with the complex FIR based approach. Results demonstrated that this method could correct the variation in real time. Therefore, they suggest its usefulness to calibrate sweep and power variation for a swept-laser. The second problem was addressed by proposing an efficient acquisition scheme and implementing required processing steps on GPU. By combining these two methods, new M-scan data can be acquired through the Alazartech card while transferring and processing the previous one. This acquisition-processing scheme allowed a series of M-scan data to be acquired and processed in near real time, showing only a little latency slightly less than acquisition time. Also, this proposed scheme was applied to perform single point M-scan, BM-scan, and volume M-scan in order to obtain additional vibrational information as well as structures at one axial point (z-axis), over two dimension (x, z), and over three dimension (x, y, z), respectively. Therefore, it can allow biologists and clinicians to obtain structural and vibrational information quickly. To tackle the last problem, the effects of additive noise and an adjacent reflector were derived in a way that they are separated from an ideal vibrational signal that does not have those effects. This separation permitted to theoretically figure out how additive noise and an adjacent reflector affect measuring vibrational amplitude and phase in frequency domain. The derived equations were verified with MATLAB simulation and then with a piezo electric element using a sweptsource OCT system. It was shown that results from derived equations matched well with those from experiments. Also, methods to reduce those effects were discussed based on the derivation. Therefore, the proposed derivation can be used not only to evaluate the performance of an OCT vibrometry system, but also to find methods to alleviate the effects. | en |
