Dai Bo, You Shaojie, Wang Kan, Long Yan, Chen Junyi, Upreti Neil, Peng Jing, Zheng Lulu, Chang Chenliang, Huang Tony Jun, Guan Yangtai, Zhuang Songlin, Zhang Dawei
Engineering Research Center of Optical Instrument and System, the Ministry of Education, Shanghai Key Laboratory of Modern Optical System, University of Shanghai for Science and Technology, Shanghai 200093, China.
Department of Neurology, Renji Hospital, School of Medicine, Shanghai Jiaotong University, Shanghai 200127, China.
Sci Adv. 2025 Jan 3;11(1):eadq2494. doi: 10.1126/sciadv.adq2494. Epub 2025 Jan 1.
Optical filtering is an indispensable part of fluorescence microscopy for selectively highlighting molecules labeled with a specific fluorophore and suppressing background noise. However, the utilization of optical filtering sets increases the complexity, size, and cost of microscopic systems, making them less suitable for multifluorescence channel, high-speed imaging. Here, we present filter-free fluorescence microscopic imaging enabled with deep learning-based digital spectral filtering. This approach allows for automatic fluorescence channel selection after image acquisition and accurate prediction of fluorescence by computing color changes due to spectral shifts with the presence of excitation scattering. Fluorescence prediction for cells and tissues labeled with various fluorophores was demonstrated under different magnification powers. The technique offers accurate identification of labeling with robust sensitivity and specificity, achieving consistent results with the reference standard. Beyond fluorescence microscopy, the deep learning-enabled spectral filtering strategy has the potential to drive the development of other biomedical applications, including cytometry and endoscopy.
光学滤波是荧光显微镜中不可或缺的一部分，用于选择性地突出显示用特定荧光团标记的分子并抑制背景噪声。然而，光学滤波组件的使用增加了显微系统的复杂性、尺寸和成本，使其不太适合多荧光通道、高速成像。在此，我们展示了基于深度学习的数字光谱滤波实现的无滤波荧光显微成像。这种方法允许在图像采集后自动选择荧光通道，并通过计算由于激发散射的存在而导致的光谱偏移引起的颜色变化来准确预测荧光。在不同放大倍数下展示了对用各种荧光团标记的细胞和组织的荧光预测。该技术以强大的灵敏度和特异性提供了对标记的准确识别，与参考标准取得了一致的结果。除了荧光显微镜外，基于深度学习的光谱滤波策略有潜力推动包括细胞计数和内窥镜检查在内的其他生物医学应用的发展。
