利用二維 MEMS 掃描振鏡於手持式光學同調斷層掃 描成像系統開發之可行性研究

Abstract

最近，隨著微型光學元件商業市場的擴大，以及在臨床應用中對於靈活性和手動操作的需求增加，手持式光學同調斷層掃描術(Opticalcoherencetomography,OCT)的開發正迅速進行。已經證實了基於多種掃描技術的手持式OCT，它們可以應用於各種廣泛的情境。低成本、便攜且易於使用的OCT系統對於在護理環境中廣泛應用至關重要，同時必須確保它們提供了足夠的成像性能，以滿足臨床檢測病灶的需求。對於即時診斷，OCT系統應該是可攜式且低成本的，同時仍保持高效能系統提供的必要靈敏度和對比。 雖然OCT系統在高解析度、非侵入性和高精確度等方面具有優勢，但它仍然受到體積、重量和成本等方面的限制，這些因素限制了可攜式且低成本的OCT系統之發展。因此，本文著重於傳統掃描設備中使用的較大的掃描振鏡(Galvanometerscanner)對於手持系統的限制，為了克服這些限制，我們決定採用微機電系統(Microelectromechanicalsystem,MEMS)掃描鏡作為系統的掃描元件。相較於傳統的掃描振鏡，MEMS掃描鏡更輕巧，能夠實現高精確度的掃描，同時佔用更少的空間。 本研究所使用的MEMS掃描鏡替代了傳統的掃描振鏡，有效縮小了手持設備的大小及複雜度。目前大多數所使用的MEMS掃描鏡為開迴路(Openloop)控制，在微型化的同時可能會犧牲掃描的準確性，為了優化MEMS掃描鏡的表現，我們結合了現場可程式化邏輯閘陣列(Fieldprogrammablegatearray,FPGA)以開迴路(Openloop)及閉迴路(Closeloop)的方式來驅動靜電MEMS掃描鏡，所架設系統的光源中心波長為1310nm，光源頻寬為100nm，掃描頻率為400kHz，最大掃描範圍為4.5mm×3.0mm，在空氣中的軸向解析度與橫向解析度分別為19.58μm和30.11μm。 本篇所使用的MEMS掃描鏡在驅動方面有開迴路及閉迴路的選擇，為了比較兩者之設計與成效，設計了一系列的驗證方法，以優化最終成像結果，包括不同入射角(22.5°、45°)的樣品端手持式緊湊性設計、利用位置感測器(Positionsensitivedetector,PSD)所接收的光點移動軌跡進行相似度及頻譜分析，以及相鄰B-scan的時間差來驗證重複掃描的穩定性，及利用哈里斯邊角偵測(HarrisCornerDetector)的方式直接從Gridtarget的OCT影像上量化失真(Distortion)程度。最後針對人類手指進行OCT血管攝影術(OCTangiography,OCTA)的成像掃描來作為開迴路及閉迴路影像穩定性的比較，可以藉此驗證是否有MEMS掃描鏡驅動控制本身的不自主運動引起的偽影，阻礙了組織微血管系統的可視化。 基於本論文所提供的測試方法，可以提供後續在MEMS掃描鏡的開迴路及閉迴路控制電路的設計上做有效的測試依據，當我們使用分析方法來評估開迴路及閉迴路驅動時，FPGA的可定制性使其成為解決閉迴路驅動問題的理想選擇。FPGA可以根據需要即時調整控制算法和參數，並且內部的DSP模塊可以有效地執行複雜的控制算法和實時數據處理，此外，極低的控制延遲以確保控制訊號能夠及時傳達到MEMS掃描鏡，使控制系統能夠快速且準確地做出反應。

Recently, with the expansion of the commercial market for micro-optical devices and the increasing demand for flexibility and manual operation in clinical applications, the development of handheld Optical Coherence Tomography (OCT) probes is rapidly progressing. Various scanning-based handheld OCT probes have been proven effective for a wide range of applications. Low-cost, portable, and user-friendly OCT systems are crucial for widespread use in healthcare environments, while ensuring that they provide sufficient imaging performance to meet the requirements of clinical lesion detection. For real-time diagnostics, OCT systems should be portable and cost-effective while maintaining the necessary sensitivity and contrast provided by high-performance systems. In this work, a MEMS scanning mirror was used to replace traditional Galvanometer (Galvo), effectively reducing the size and complexity of handheld devices. It was combined with a Field-Programmable Gate Array (FPGA) for both open-loop and closed-loop control of the electrostatic MEMS scanning mirror. The system utilized a light source with a center wavelength of 1310nm, a bandwidth of 100nm, a scanning frequency of 400 kHz, and a maximum scanning range of 4.5 mm×3.0 mm. The axial and lateral resolutions in air were 19.58 μm and 30.11 μm , respectively. For MEMS drive control, there was a choice between open-loop and closed-loop control. To compare the design and performance of both approaches, a series of validation methods were designed to optimize the final imaging results. This included a compact handheld design with different incident angles (22.5° and 45°) at the sample end, similarity and spectrum analysis of the trajectory of the received light spots using a Position-Sensitive Detector (PSD), verification of the stability of repeated scans using the time difference between adjacent B-scans, and quantifying distortion directly from OCT images of a grid target using the Harris Corner Detector. Finally, OCTA imaging scans of the human finger were performed to compare the stability of open-loop and closed-loop images, aiming to verify whether any artifacts caused by involuntary motion of the MEMS scanning mirror control hindered the visualization of the microvascular system in tissues. Based on the testing methods provided in this thesis, effective testing criteria can be established for the design of open-loop and closed-loop control circuits for MEMS scanning mirrors. When evaluating open-loop and closed-loop drive control using analytical methods, the FPGA''s customizability makes it an ideal choice for addressing closed-loop drive control issues. The FPGA can adjust control algorithms and parameters in real-time as needed, and its internal DSP modules can effectively execute complex control algorithms and real-time data processing. Additionally, the extremely low control latency ensures that control signals are delivered to the MEMS scanning mirror in a timely manner, allowing the control system to respond quickly and accurately.