Breakthrough in deep tissue super-resolution imaging via confocal² spinning-disk ISM

Existing super-resolution techniques still face considerable challenges in complex deep tissue environments. For example, Stimulated Emission Depletion (STED) microscopy relies on a doughnut-shaped depletion beam, which is highly susceptible to distortion by tissue scattering, leading to a significant degradation in image quality at depth. Structured Illumination Microscopy (SIM), which depends on stripe illumination patterns, is easily disrupted in scattering media, resulting in reconstruction artifacts. Single-Molecule Localization Microscopy (SMLM) suffers from reduced localization accuracy due to background fluorescence and scattering interference.

 

Image Scanning Microscopy (ISM), as an evolution of confocal microscopy, preserves excellent optical sectioning capability while enhancing resolution to nearly twice the diffraction limit through pixel reassignment (PR) and deconvolution. In 2024, the research team proposed Multi-Confocal Image Scanning Microscopy (MC-ISM, Natl Sci Rev, 2024), which addresses the trade-offs between spatial and temporal resolution inherent in conventional ISM approaches. Nevertheless, super-resolution imaging in deep tissue still suffers from issues such as background interference, limited imaging depth, and insufficient fidelity.

 

In a new paper published in Light: Science & Applications, a team of scientists, led by Professor Peng Xi from Department of Biomedical Engineering, College of Future Technology, Peking University, China, and co-workers have developed Confocal² Spinning-Disk Image Scanning Microscopy (C²SD-ISM). The system integrates a spinning-disk confocal microscope, which physically removes out-of-focus signals, establishing the first confocal level. Additionally, a digital micromirror device (DMD) enables sparse multifocal illumination, while a dynamic pinhole array pixel reassignment (DPA-PR) algorithm for ISM super-resolution reconstruction, forming the second confocal level. Accordingly, the technique is termed “dual-confocal” (Confocal²), as illustrated in the optical schematic in Fig. 1a. Under thick tissue conditions and dense multifocal excitation patterns, the system maintains high visibility of excitation foci (Fig. 1b–d). Compared to conventional multifocal structured illumination microscopy (MSIM), it requires approximately six times fewer raw images for reconstruction.

 

Conventional ISM reconstruction is based on PR, which assumes that the excitation and detection PSFs are identical and ideal Gaussian profiles. However, this simplification neglects practical factors such as the Stokes shift and system aberrations. To address these limitations, the research team developed the DPA-PR algorithm. This method constructs a 5×5 virtual detector array (VDA) and extracts multiple offset sub-images from the raw image stack. The spatial offsets are estimated via phase cross-correlation, enabling high-fidelity image reconstruction (Fig. 2). The reconstructed images exhibit a linear correlation coefficient of up to 92% with the original confocal images, demonstrating high-fidelity super-resolution. A lateral resolution of 144 nm was achieved in tissue specimens.

 

In validation experiments using strongly scattering mouse kidney sections, C²SD-ISM demonstrated robust performance in resolving complex structural regions. Compared to computational background removal methods, the combination of spinning-disk-based defocus removal followed by DPA-PR reconstruction not only preserved the structural continuity of weak-signal regions but also maintained a linear relationship between the reconstructed signal and the original intensity distribution. This enables a more faithful representation of the underlying sample structure.

 

C²SD-ISM successfully performed 3D imaging over a 66.5 × 66.5 × 12 μm volume in mouse kidney section, with axial steps of 150 nm and lateral/axial resolutions of 162 nm and 351 nm, respectively (Fig. 3a-3c). The research team also performed volumetric mosaic imaging of the EGFP-labeled zebrafish vasculature, covering a field of view of 2.91 mm × 1.26 mm × 0.18 mm. Compared to conventional confocal imaging, C²SD-ISM markedly enhanced spatial resolution, revealing fine vascular structures that were difficult to discern in standard confocal images and presenting a significantly clearer vascular morphology (Fig. 3d).

 

By loading striped pattern masks onto the DMD, the system enables projection-based Structured Illumination Microscopy (SIM). The physical removal of out-of-focus signals by the spinning disk significantly enhances the modulation contrast of illumination stripes in thick tissue samples. Under the SIM mode, the system achieved 3D imaging of fungal samples over a volume of 66.5 × 66.5 × 50 μm, surpassing the depth limitations of conventional SIM and achieving an approximate 1.68-fold resolution enhancement (Fig. 3e-3g).

 

C²SD-ISM achieves an optimal balance among resolution, imaging depth, and image fidelity through its dual-confocal architecture and adaptive reconstruction algorithm. By overcoming the limitations of conventional confocal and super-resolution techniques in thick-tissue imaging, the system demonstrates high throughput, multicolor compatibility, and extendibility to structured illumination imaging—highlighting its flexibility and practical utility. In the future, by integrating deep learning-based denoising and adaptive optical correction, C²SD-ISM is expected to further enhance imaging performance and scale up to deeper and larger-volume 3D imaging applications. Moreover, its DMD programmability aligns well with adaptive illumination strategies, offering the potential for intelligent, low-phototoxicity imaging of dynamic cellular processes. With its versatile capabilities, C²SD-ISM shows great promise for both cellular and tissue-scale imaging, providing a powerful tool for advancing biological research.

 

All essential resources used in this study have been made openly available, including simulation code for artifact-free spinning-disk confocal imaging, mask design files for disk fabrication, super-resolution reconstruction code for multifocal excitation data based on the DPA-PR algorithm, simulation tools for comparing widefield and confocal 3D imaging based on light field propagation, and hardware control software (https://github.com/Chauncey-Leung/C2SD-ISM). We believe that open-sourcing these components will facilitate broader scientific collaboration and the dissemination of this technology.

 

Part of this technology has been successfully commercialized as the Nova-SD spinning-disk confocal system, which offers a native lateral resolution of 230 nm, imaging speeds up to 2000 fps, seven-channel excitation, and a maximum field of view of 25 mm

---

---