1. Background
Three-dimensional (3D) observation of dynamic biological samples holds profound significance in both fundamental research and technological applications. Taking marine ecosystems as an example, analyzing the 3D movement patterns of plankton (e.g., copepods, ciliates) is crucial for understanding changes in marine ecosystems and the impact of environmental variations on marine plankton.
Light-field microscopy, by capturing four-dimensional light-field information (two-dimensional spatial + two-dimensional angular data) and combining computational reconstruction algorithms, enables non-scanning 3D imaging. Observing dynamic biological processes with light-field microscopy requires the introduction of a fifth dimension—temporal resolution—to acquire five-dimensional data (two-dimensional spatial + two-dimensional angular + one-dimensional temporal). Current mainstream light-field microscopy technologies face notable limitations in achieving five-dimensional data acquisition (Figure 1): micro-lens array-based light-field microscopy (MLA-based light-field microscopy) sacrifices spatial resolution for high temporal resolution; image-coding-based light-field microscopy, while offering high angular resolution, suffers from a trade-off between temporal and spatial resolution; and aperture-coding-based light-field microscopy, despite its high spatial resolution, struggles with the mutual constraints of temporal and angular resolution. Spatial resolution is a key technical metric for microscopy, and how to enhance temporal resolution without compromising spatial resolution has become a critical issue in light-field microscopy.
Figure 1. Comparison of trade-off strategies in different light-field microscopy techniques. (a) Micro-lens array-based light-field microscopy; (b) Image-coding-based light-field microscopy; (c) Aperture-coding-based light-field microscopy.
2. Research Content
Recently, a research team led by Professor Jingang Zhong from Jinan University, in collaboration with Professor Jianping Li's group at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, published a paper titled "Differential High-Speed Aperture-Coding Light Field Microscopy for Dynamic Sample Observation with Enhanced Contrast" in Advanced Imaging. This study introduces a novel differential high-speed aperture-coding light field microscopy technique. By optimizing aperture coding, image acquisition rate, and light field reconstruction algorithms, the technique significantly enhances the temporal resolution and imaging contrast of light field imaging without compromising spatial resolution. This advancement provides a new solution for the observation of dynamic samples.
The optical design of the imaging system is illustrated in Figure 2. Figure 2(a) shows the main imaging path, where a spatial light modulator (SLM) is placed at the back focal plane of a Fourier lens to encode the Fourier spectrum. A camera captures the modulated 2D intensity image (spatial information) at the image plane, with each pixel acting as an independent single-pixel detector. The study employs a single-pixel imaging method to reconstruct the spectral image (angular information). Single-pixel imaging follows the Helmholtz reciprocity principle, where the SLM functions as a virtual camera in the dual optical path (Figure 2(b)), and the camera acts as a virtual SLM. The dual image reconstructed using single-pixel imaging corresponds to the image captured by the virtual camera in the dual path. Thus, the system simultaneously utilizes the real camera to acquire spatial information and the virtual camera to obtain angular information, effectively resolving the mutual constraints between angular and spatial resolution.
Figure 2. Schematic diagram of the differential high-speed aperture-coded light-field microscopy. (a) Main imaging path; (b) Dual imaging path.**
3. Technical Breakthroughs and Innovations
The highlights of the paper are summarized in the following three aspects:
(1) High-Speed Aperture Encoding and Image Acquisition, Breaking Spatiotemporal Resolution Limits
Traditional aperture-coded light-field microscopy faces a trade-off between spatial and temporal resolution. To address this challenge, the study employs a high-speed SLM to rapidly encode the Fourier spectrum at the Fourier plane, while a high-speed camera captures the encoded images. This synergistic mechanism of high-speed encoding and image acquisition significantly enhances the temporal resolution of light-field imaging, enabling real-time observation of dynamic processes.
(2) Differential Encoding Strategy, Effectively Enhancing Light-Field Image Contrast
To mitigate the unavoidable DC background noise in aperture-coded light-field microscopy, the study proposes a differential encoding strategy. By loading a set of phase-shifted Fourier basis patterns onto the SLM and utilizing a differential single-pixel imaging algorithm, the DC background noise recorded by the camera is effectively eliminated, thereby improving the contrast of light-field images. Figure 3 compares the light-field images reconstructed using the differential encoding strategy and those without it (based on S-matrix aperture-coded light-field microscopy). Experimental results demonstrate that the contrast of images reconstructed with the differential encoding strategy is twice that of S-matrix-based methods, highlighting the advantages of this strategy in enhancing image quality.
Figure 3. Performance comparison between the proposed method and S-matrix-based aperture-coded light-field microscopy.
(3) Undersampling Strategy, Effectively Improving Temporal Resolution
Leveraging the energy accumulation characteristics of spectral images, an undersampling strategy is proposed to enhance the temporal resolution of light-field imaging without sacrificing spatial resolution or depth of field. Experimental results (Figure 4) show that at a 60% sampling rate, the reconstructed light-field images exhibit comparable depth of field and spatial resolution to those at a 100% sampling rate. Notably, the proposed method does not require complex iterative optimization algorithms during undersampling, simplifying the computational process and improving imaging efficiency, thereby enabling dynamic observation.
Figure 4. Performance comparison of the proposed method at different sampling rates.
4. Applications
To validate the effectiveness and feasibility of the proposed method, it was applied to imaging experiments on dynamic copepod samples. The experiments used an objective lens with a numerical aperture of 0.3 and a magnification of 10×, reconstructing a light-field volume of 1518 μm × 1172 μm × 600 μm. The frame rates of the SLM and camera were set to 1340 Hz. Figure 5 compares the imaging results of wide-field microscopy and the proposed method at different sampling rates.
Experimental results show that at a 100% sampling rate, 53 measurements are required to reconstruct the light field, achieving a temporal resolution of 25 Hz. When the sampling rate is reduced to 60%, the number of measurements decreases to 32, and the temporal resolution significantly improves to 41 Hz, with almost no degradation in imaging quality. This finding confirms that by appropriately reducing the sampling rate, the temporal resolution of light-field imaging can be effectively enhanced while maintaining imaging quality, providing a new technical approach for observing dynamic biological samples.
Figure 5. Imaging results of dynamic copepod samples. (a1-e1) Traditional wide-field microscopy; (a2-e2) Results obtained using the proposed method at a 100% sampling rate; (a3-e3) Results obtained using the proposed method at a 60% sampling rate.
5. Conclusions and Future Perspectives
This paper presents a differential high-speed aperture-coding light field microscopy method. By employing a differential coding strategy, the proposed method achieves a twofold improvement in the contrast of light field imaging results compared to conventional S-matrix-based aperture-coding light field microscopy. Furthermore, by utilizing a high-speed spatial light modulator and a high-speed camera in conjunction with an undersampling mechanism, the temporal resolution of aperture-coding light field microscopy is significantly enhanced without compromising spatial resolution. Through further optimization of the coding scheme and the use of spatial light modulators and cameras with higher frame rates, the temporal resolution of light field imaging is expected to be further improved. The method proposed in this study offers a novel solution for the observation of dynamic samples.
