Super-resolving lensless microscopy via optimized multi-depth fractional Talbot modulation

High-throughput fluorescence imaging is central to fields ranging from immunomics and functional genomics to neuroscience. Lensless microscopy offers an attractive route to meeting such demands through its superior space-bandwidth product (SBP). This framework achieves micrometer-scale resolution across a wide field of view, spanning millimeters to centimeters. However, the resolution of lensless fluorescence microscopy is subject to multifaceted physical and hardware constraints that make it difficult to achieve high-fidelity imaging at the scale required for biological observation. We present spatially encoded high-resolution lensless imaging (Se-HLi), a computational framework that enhances the resolution of compact lensless architectures. Se-HLi uses a movable diffraction grating to generate multi-depth fractional Talbot modulation, thereby encoding high-spatial-frequency information into measurable sensor patterns. A physics-informed spatial modulator aware resolution-improvement transformer then recovers super-resolved details from these measurements. Through differentiable end-to-end optimization of both grating positions and network parameters, Se-HLi improves the system resolution from 5.6 to 3.1 μm (full width at half maximum), while maintaining a minimal hardware footprint. Experimental results show a nearly twofold improvement in resolution, accompanied by a more than threefold increase in SBP. The Se-HLi framework offers a route toward compact, high-throughput platforms for wide-field fluorescence imaging.