Instant mapping of digestive morphology via spatio-angular disparity-resolved endoscopy

Accurate volumetric information is critical for early detection, diagnosis, and precise resection of gastrointestinal lesions, thereby preventing malignant transformation. However, conventional 2D endoscopy lacks objective 3D spatial data and relies on subjective interpretation, increasing the risk of missed lesions and surgical complications. The inaccurate estimation of lesion size exacerbates diagnostic errors and impairs decision-making, potentially leading to suboptimal clinical outcomes. To address these limitations, we developed High-resolution Endoscopy with Resolved spatio-Angular disparity for Layered Depth Sensing (HERALDS), which integrates four miniaturized imaging units into a compact probe (∼5.8 × 5.8 × 5.0 mm), enabling spatio-angular data acquisition in confined anatomical spaces. Leveraging an innovative computational pipeline, the system reconstructed volumetric tissue data from a 1.4 cm³ region in approximately 1 s, achieving a lateral resolution of 19.69 μm and enabling accurate quantification of lesion size. Compared with conventional 3D Richardson-Lucy deconvolution, our method maintains lateral resolution while providing an order-of-magnitude acceleration and eliminating defocus-related artifacts, thus improving the efficiency of endoscopic procedures and potentially alleviating physicians’ workload and patients’ exposure to procedure-associated discomfort. Through imaging 3D-printed models and in vivo animal subjects, we demonstrated the functionality of HERALD, which enables the real-time detection of structural abnormalities that are often missed by conventional techniques. These results underscore HERALD'S clinical potential in improving lesion detection, diagnostic precision, surgical guidance, and overall diagnostic efficiency.