以樂高積木構建之低成本層光顯微鏡開發與其於三維生物影像化之應用

Abstract

商用層光顯微鏡因硬體造價與光學對齊門檻，限制了其於一般生物實驗室與教育環境之普及。本研究以開源 LEMOLish 架構為基礎，開發一套低成本、模組化且具自動化掃描功能之桌面型層光顯微成像平台。系統保留樂高積木結構框架，並導入 3D 列印固定式狹縫、實心玻璃柱、對稱式雙向照明、樹莓派相機模組、EV3 馬達控制及網頁操作介面，以改善光學整形穩定性、樣本照明均勻性與影像擷取效率。

實驗結果顯示，改良型狹縫可降低局部異常高亮與不規則灰階變化。雙向照明則可補償單側照明造成的方向性亮度衰減與遮蔽。以自製斜邊靶標量測所得之感測器平面 MTF50、MTF20 與 MTF10 分別為 102 lp/mm、225 lp/mm 與 324 lp/mm，換算至樣本平面後，其等效半週期尺寸約為 98.39 μm、44.58 μm 與 30.92 μm。影像預處理後，各組影像之 CNR 均明顯提升，CII 都大於 1，顯示其對比提升。兩次重複掃描體數據於 z、y、x 方向的相對位移約為 50 μm、81 μm 與 0 μm，配準後峰值相關係數為 0.7193。三維重建結果可呈現透明化魷魚樣本之整體輪廓、眼球外廓與腕足等大尺度組織結構。Image Decorrelation Analysis 所得 XY 截面面內有效解析度約為 50 μm，但不代表整體三維解析度。

本平台硬體建置成本約為 2,760 美元，驗證了以樂高積木、3D 列印元件、消費級相機與開源控制程式建構公分級透明樣本三維成像系統之可行性。此系統雖仍受限於層光厚度、Z 軸取樣及細節辨識能力，但可作為低成本三維生物成像與跨領域 STEM 教育之實作平台。

Commercial light-sheet microscopes (LSMs) are often constrained by high hardware costs and complex optical alignment procedures, which limits their widespread adoption in general biological laboratories and educational environments. To overcome these physical and financial barriers, this study developed a modular, desktop-class LSM platform based on the open-source LEMOLish architecture. While retaining the LEGO-based structural framework, the platform incorporates 3D-printed fixed slits, solid glass cylinder lenses, symmetrical dual-sided illumination, a Raspberry Pi camera module, EV3 motor control, and a web-based user interface to enhance optical beam-shaping stability, illumination uniformity, and image acquisition efficiency.

Experimental results demonstrate that the modified slits effectively eliminate localized intensity hotspots and irregular grayscale variations. Furthermore, symmetrical dual-sided illumination compensates for directional attenuation and shadowing artifacts inherent in single-sided illumination. Using a custom slanted-edge target, sensor-plane MTF50, MTF20, and MTF10 were measured at 102 lp/mm, 225 lp/mm, and 324 lp/mm, corresponding to equivalent sample-plane half-period resolution limits of approximately 98.39 µm, 44.58 µm, and 30.92 µm, respectively. Following image preprocessing, the Contrast-to-Noise Ratio (CNR) was significantly improved across all datasets, with a Contrast Improvement Index (CII) consistently exceeding 1.0. Volumetric repeatability tests revealed relative displacements of approximately 50 µm, 81 µm, and 0 µm along the Z-, Y -, and X-axes between repeat scans, yielding a post-registration peak correlation coefficient of 0.7193. Three-dimensional reconstructions clearly resolved macro-structural features of optically cleared squid specimens, including overall morphology, eye contours, and arms. Image Decorrelation Analysis indicated an in-plane (XY) effective resolution of approximately 50 µm.

Constructed at a hardware cost of approximately $2,760 USD, this platform validates the feasibility of building a 3D imaging system for centimeter-scale cleared specimens using LEGO bricks, 3D-printed components, consumer-grade cameras, and open-source control software. Despite current constraints in light-sheet thickness, Z-axis sampling, and fine feature resolution, the proposed system serves as a cost-effective practical platform for 3D biological imaging and interdisciplinary STEM education.