Department of Interdisciplinary Research & Development, Graduate School of Medical Science, Kyoto Prefectural University of Medicine (KPUM), Kyoto, Japan.
Department of Pathology and Applied Neurobiology, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto, Japan.
Sci Rep. 2019 Jan 24;9(1):633. doi: 10.1038/s41598-018-37241-y.
Due to a growing demand for a viable label-free observation method in the biomedical field, many techniques, such as quantitative phase imaging and Raman spectroscopy, have been studied, and a complementary approach, hyperspectral imaging, has also been introduced. We developed a high-speed hyperspectral imaging microscopy imaging method with commercially available apparatus, employing a liquid crystal tunable bandpass filter combined with a pixel-wise machine learning classification. Next, we evaluated the feasibility of the application of this method for stem cell research utilizing neural stem cells. Employing this microscopy method, with a 562 × 562 μm field of view, 2048 × 2048 pixel resolution images containing 63 wavelength pixel-wise spectra could be obtained in 30 seconds. The neural stem cells were differentiated into neurons and astroglia (glia), and a four-class cell classification evaluation (including neuronal cell body, glial cell body, process and extracellular region) was conducted under co-cultured conditions. As a result, an average of 88% of the objects of interest were correctly classified, with an average precision of 94%, and more than 99% of the extracellular pixels were correctly segregated. These results indicated that the proposed hyperspectral imaging microscopy is feasible as a label-free observation method for stem cell research.
由于生物医学领域对可行的无标记观察方法的需求不断增长，许多技术（如定量相位成像和拉曼光谱学）已经得到研究，并且还引入了一种互补的方法，即高光谱成像。我们使用商业上可用的仪器开发了一种高速高光谱成像显微镜成像方法，采用液晶可调谐带通滤波器与逐像素机器学习分类相结合。接下来，我们评估了该方法在利用神经干细胞进行干细胞研究中的应用的可行性。使用这种显微镜方法，可以在 30 秒内获得包含 63 个波长逐像素光谱的 562×562μm 视场、2048×2048 像素分辨率的图像。神经干细胞分化为神经元和神经胶质（胶质），在共培养条件下对四类细胞分类评估（包括神经元细胞体、神经胶质细胞体、突起和细胞外区）进行了评估。结果，平均 88%的感兴趣对象被正确分类，平均精度为 94%，超过 99%的细胞外像素被正确分离。这些结果表明，所提出的高光谱成像显微镜作为一种无标记观察方法，可用于干细胞研究。
