Hybrid iterative ptychographic super-resolution microscopy

Achieving high resolution alongside low photodamage in a simple system remains a critical pursuit in biomedical imaging. Although conventional image scanning microscopy (ISM) is widely used due to its simple system configuration, low photodamage, excellent optical sectioning, and high signal-to-noise ratio (SNR), further breaking through resolution limits has traditionally required high-power depletion lasers, which inevitably leads to a severe dilemma where elevated resolution is paired with heavy photodamage. Furthermore, conventional ISM reconstruction algorithms are predominantly built upon linear shift invariance models, restricting their theoretical capability to a frequency domain concatenation of optical transfer functions (OTF). This fundamental limitation prevents the sufficient extraction of high frequency information and restricts bandwidth extrapolation, making it difficult to resolve finer subcellular details under gentle and low-photodamage conditions.

In a new study published in PhotoniX, researchers from the College of Optical Science and Engineering at Zhejiang University introduce hybrid iterative ptychographic super-resolution microscopy (HiM). The work demonstrates how HiM bypasses the traditional requirement for high-power depletion lasers to enhance resolution, successfully achieving a spatial resolution of approximately 60 nm under gentle, low-damage conditions. By seamlessly integrating advanced optical hardware with computational reconstruction strategies, this approach establishes a powerful new framework for high-performance super-resolution imaging.

At the hardware level, the researchers introduced a honeycomb array detector architecture integrated with microlens arrays (MLA), overcoming the geometric fill-factor bottlenecks caused by inter-element gaps in conventional detector arrays. This design increased the average photon collection efficiency of HiM by approximately 33% compared to conventional ISM, laying a high-quality data foundation for downstream reconstruction algorithms. At the reconstruction algorithm level, driven by a virtual modulation mechanism via pure-cosine digital masks and combined with a hybrid iterative reconstruction strategy alternating between virtual demodulation and sparse joint deconvolution, the framework strictly honors underlying physical models while enabling effective frequency extrapolation and the accurate reconstruction of super-resolution images.

Demonstrated on fluorescent line pairs as well as diverse biological samples including mitochondrial outer membranes, microtubules, actin filaments, nuclear pore complexes, and retraction fibers, HiM achieved an approximate spatial resolution of 60 nm under gentle conditions free of extra high-power depletion light. It substantially mitigated phototoxicity and photobleaching while delivering an apparent resolving capability comparable to stimulated emission depletion microscopy (STED), successfully balancing high photon collection efficiency, high spatial resolution, and low photodamage. Furthermore, the research team seamlessly extended and integrated this technological framework into advanced optical modalities such as low-power STED and fluorescence lifetime imaging microscopy (FLIM), validating its powerful scalability in multi-modal super-resolution and functional bio-imaging.

Building upon these advances, this work highlights the effectiveness of deeply fusing hardware architecture with computational reconstruction to push the performance boundaries of super-resolution microscopy, offering a practical pathway for future developments in the field of super-resolution. Looking ahead, the introduction of lightweight deep learning networks and adaptive optical aberration correction is expected to bring further improvements to the HiM framework, enhancing its processing speed and environmental adaptability for deep, complex tissue imaging.