TL;DR
This paper introduces a novel spectral Confocal Spinning Disk Image Scanning Microscopy (CSD-ISM) system that achieves simultaneous multi-color, high-resolution imaging. By integrating a custom Amici prism, the system captures spectral information alongside spatial data, significantly reducing acquisition time and eliminating temporal artifacts common in sequential multi-color imaging.
What Problem Does This Paper Address?
Traditional fluorescence microscopy, while powerful for visualizing biological samples, often struggles to resolve fine details and simultaneously capture information across multiple colors. Image Scanning Microscopy (ISM), particularly when combined with Confocal Spinning Disks (CSD), has improved spatial resolution beyond the diffraction limit. However, acquiring multi-color images with CSD-ISM remains a bottleneck. The standard approach involves acquiring separate images for each color sequentially, which is time-consuming and can lead to "spatio-temporal chromatic artifacts" where different colors are captured at slightly different times, distorting dynamic processes. This paper aims to overcome this by enabling simultaneous acquisition of both high spatial resolution and multi-color information.
Methods
The core innovation lies in the integration of a custom linear Amici prism into the CSD-ISM optical path. This prism disperses the emitted fluorescence from each confocal spot into its spectral components along the horizontal axis, effectively encoding color information directly onto the camera sensor. This spectral encoding is performed simultaneously with the ISM's spatial scanning.
For image processing, the authors developed a GPU-compatible Python pipeline. This pipeline first calibrates the system's spectral response using known fluorescent samples. Then, it uses a linear regression approach to decompose the spectral signatures captured by each spot into their constituent colors. This decomposition is performed on regions of interest (ROIs) around the localized illumination spots. Finally, the spatially reassigned spectral information is reconstructed into a high-resolution, multi-color image. A Digital Signal Processor (DSP) is used for precise synchronization of the spinning disk, laser, and camera, which is crucial for ISM.
Datasets
The paper utilizes several datasets for characterization and validation:
- Fluorescent Beads:
- 3-color mixture of 170 nm diameter fluorescent beads with emission maxima at 515 nm (blue), 560 nm (green), and 660 nm (red) for system validation.
- 4 µm multi-colored fluorescent beads (blue, green, orange, deep red) for image analysis pipeline validation.
- Biological Samples:
- Neuron cells expressing endogenous eGFP and labeled with antibodies for G3BP (CF568) and synaptotagmin-1 (CF647), along with DAPI for nuclei staining. This dataset was used to demonstrate the system's capability in a biologically relevant context, showcasing improved resolution and contrast of cellular components.
Key Insights and Ideas
- Spectral Encoding as a Solution: The clever use of an Amici prism to spectrally disperse light before it hits the camera is a key insight. This allows for simultaneous color capture without the need for sequential filter changes.
- Bridging Spatial and Spectral Information: The system effectively links the spatial information captured by ISM with the spectral information encoded by the prism, creating a unified high-resolution, multi-color dataset.
- Cost-Effective DSP Implementation: Utilizing a DSP instead of more expensive FPGAs for synchronization makes the system more accessible.
- Linear Regression for Color Decomposition: The use of linear regression for spectral-to-color conversion is computationally efficient and allows for rapid processing of millions of ROIs.
- Preserving Resolution During Color Assignment: While spectral dispersion can slightly impact resolution, the authors demonstrate that their approach, combined with ISM reassignment and deconvolution, largely preserves the super-resolution gains.
Key Results
- Resolution Enhancement: The system demonstrated a 1.74-fold resolution improvement over the diffraction limit for deconvolved color ISM images compared to the sum of raw spectral images.
- Speed Improvement: The spectral CSD-ISM system achieved three times faster acquisition compared to standard CSD-ISM for multi-color imaging.
- Accurate Color Classification: The system showed accurate color classification of fluorescent beads.
- Biological Application: Imaging of neuron cells revealed improved resolution and contrast of four distinctly labeled cellular components, enabling better visualization of subcellular structures.
Limitations and Future Work
- Parallax Effect and Resolution: The paper acknowledges a slight decrease in lateral resolution during color assignment due to the limited accounting for parallax effects with a single Taylor approximation term. While ISM reconstruction and deconvolution help overcome this, further improvements could be made by refining the regression analysis to better account for parallax.
- Non-Overlapping Spectral PSFs: A limitation is the requirement for non-overlapping spectral Point Spread Functions (PSFs) for accurate color decomposition. This means fluorophores must have sufficiently separated emission spectra. The authors suggest that future work using machine learning could potentially overcome this limitation by exploiting subtler spectral nuances.
- Signal-to-Noise Ratio (SNR) and Frame Rate: Some artifacts were observed in biological samples, attributed to low SNR and frame rate limitations. Future work will focus on optimizing imaging parameters and hardware alignment to improve SNR and reduce artifacts.
- Future Directions: The authors suggest expanding the system's capabilities for 3D imaging and dynamic live-cell applications. They also envision an "on-the-fly" process where color decomposition is calculated alongside spectral image acquisition, further enhancing usability.
Summary updated