Improvement of depth-sensitive optical spectroscopy and imaging with wavefront shaping and radiofrequency tagging

• Main Author: Hsieh, Chao-Mao
• Other Authors: Duan Hongwei
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Science::Medicine::Biomedical engineering; Science::Physics::Optics and light
• Online Access: https://hdl.handle.net/10356/168690
• Institution: Nanyang Technological University

The dissertation describes a series of studies developing depth-sensitive optical spectroscopy assisted by wavefront shaping and radiofrequency tagging. Depth-sensitive optical spectroscopy preferentially detects optical spectra from different depths in layered samples, which plays a crucial role in many applications, such as the optical diagnosis of epithelial precancer and cancer. For depth-sensitive optical measurements, multiple light scattering effects and the requirement of axial scanning cause limit the performance of related techniques. First, multiple light scattering in a highly scattering sample, such as a tissue, significantly degrades the depth sensitivity to a subsurface target layer. Second, the performance of enhancement techniques such as wavefront shaping can be unstable in dynamic samples, for example, during in vivo measurements of tissues with blood flow right beneath. Third, axial scanning achieved by mechanical motion is time-consuming, especially for an extended depth range. To address these issues, three techniques, including “feedback-based wavefront shaping with lock-in method,” “FPGA-based AOD modulation,” and “radio-frequency tagging for depth-resolved imaging,” are developed to improve depth-sensitive optical spectroscopy. In the first study in this dissertation (Chapter 3), spatial and spectral filtering can improve the performance of feedback-based wavefront shaping. The lack of intrinsic guide stars in tissues or tissue-like samples often leads to the poor enhancement of depth-sensitive Raman/fluorescence measurements (~20% in the past literature) from a target layer due to the contributions from the overlaying non-target layers. In this study, we demonstrate that spatial and spectral filtering can significantly improve the performance of depth-sensitive fluorescence spectroscopy assisted by feedback-based wavefront shaping (using an SLM) in tissue-like scattering phantoms. The two filtering techniques work by effectively increasing the relative contribution from the target layer to the feedback signal during wavefront optimization through spatially and spectrally rejecting off-target fluorescence light, essentially similar to the role of time or coherence gating. To speed up wavefront shaping for real applications, we explore using an AOD as a phase modulator instead of SLM in the second study (Chapters 4 and 5). The high refreshing rate and optical frequency shifts of the AOD provide unique advantages for wavefront shaping. It is well known that the signal-to-noise ratio is essential for feedback-based wavefront shaping. The technique can fail for low optical signals such as fluorescence and Raman signals or in a reflection setup because the trend in weak feedback signals can be easily overwhelmed by noise. To address this challenge, we develop a technique based on a single acousto-optic deflector (AOD) (Chapters 4 and 5) to create a signal with a selected beat frequency from optical signals that can serve as feedback, in which the phase distribution of various radio frequency components of the driving signal for the AOD is optimized for wavefront shaping. By shifting incident light frequency with the AOD, the feedback signal at a selected beat frequency can be measured with a high signal-to-noise ratio by a lock-in amplifier, thus enhancing weak target signals through highly scattering media. The lock-in beat-frequency detection method can significantly improve fluorescence imaging and Raman spectral measurements in a reflection setup and thus could be potentially used for in vivo measurements. Feedback-based wavefront shaping can perform poorly for dynamic samples due to short correlation time. So far, most proposed methods for high-speed wavefront shaping can only enhance the intensity of coherent light but not incoherent light. Therefore, it is necessary to develop high-speed wavefront shaping to directly enhance incoherent light, such as fluorescence, which is essential in extending wavefront shaping to biomedical applications in which fluorescence spectroscopy/imaging is broadly adopted. For this purpose, we develop a technique based on a single acousto-optic deflector (AOD) with field-programmable gate array (FPGA) acceleration for spatiotemporal focusing within milliseconds (Chapter 6). With the digital time gating of the feedback signal, the spatiotemporal focusing of laser light with high contrast can be formed behind dynamic scattering media in milliseconds resulting in significant fluorescence enhancement. Furthermore, FPGA-based wavefront shaping is shown to effectively enhance fluorescence directly behind dynamic samples with short correlation times. The final study (Chapter 7) is to improve the speed of depth-sensitive measurements by generating multiple foci along the axial dimension, with each depth encoded by a unique radio frequency. With remote focusing and multiple focal spots at different depths, each with a unique radiofrequency beat signal generated using a tilted mirror and an AOD, respectively, signals from different depths in a sample can be simultaneously measured with a fast single-pixel detector at high speed. Because the depth information is encoded in the RF signal, we can reconstruct depth-relevant information from the data measured by the single-pixel detector. This technique dramatically increases the speed of depth-sensitive spectroscopy without involving mechanical movement. This method can complement the current methods for sequential depth-dependent measurements, involving a piezo actuator, an electrically tunable lens, and a TAG lens, by increasing effective exposure time and improving the overall measurement throughput.