Mid-infrared microscopy for label-free digital staining: initial clinical assessment

Abstract

Infrared (IR) microscopy shows substantial potential for label-free tissue imaging in anatomic pathology, providing rich biochemical contrast. However, existing IR imaging technologies are constrained by slow acquisition speeds and limited spatial resolution. Here, we present a rapid, large-field bimodal imaging platform that integrates conventional brightfield microscopy with a lensless IR imaging scanner, enabling whole-slide IR image stack acquisition in minutes. Using a dedicated deep learning model, we implement an optical H&E staining strategy based on subcellular morpho-spectral fingerprinting. This approach achieves high-resolution visualization of tissue architecture with an effective spatial resolution of 500 nm, without chemical staining. Quantitative metrics, including PSNR (∼24), MS-SSIM (∼0.82), and LPIPS (∼0.22), validate the model's ability to accurately reproduce both the contrast and morphology of cellular structures. Additionally, an initial clinical evaluation on 110 regions of interest within 5 tissue sections demonstrates equivalence between our digital IR-based AI staining and conventional chemical staining, both in image quality and Gleason grading. Together, these results suggest that this IR-based virtual staining approach could provide a fast, chemical-free alternative for anatomic pathology workflows.