Computational X-Ray Technique Enables High-Resolution, Low-Dose Radiography

This is especially critical in pediatrics, pregnancy, and for patients who require repeated scans. Safer, lower‑dose imaging could expand access while limiting cumulative exposure risks. Researchers have now demonstrated a computational X-ray method designed to preserve high image detail while using a very small fraction of the usual photon dose.

Researchers at the Shanghai Advanced Research Institute, Chinese Academy of Sciences, report an ultra‑low‑light X-ray ghost imaging approach that targets dose reduction without sacrificing resolution. The method was described in Optica, a journal of Optica Publishing Group. The proof‑of‑concept work shows that high‑resolution images can be reconstructed with far fewer X-ray photons than conventional radiography typically requires.

Ultra-low-light X-ray ghost imaging recovers image information by correlating two beams. A weak beam passes through the object, while a stronger reference beam carries a random pattern that serves as a template; correlating the two signals reconstructs the image. The team used highly uniform, strongly correlated X-ray crystal splitting to acquire synchronous signals from both beams, paired detectors optimized for faint signals and fine spatial detail, and a synthetic-aperture reconstruction algorithm that builds a full high-resolution image from dozens of measurements.

In controlled comparisons on the BL09B test beamline at the Shanghai Synchrotron Radiation Facility, the method produced nearly 2-megapixel ghost images while using just 0.48% of the photons typically required. It generated 1992 × 944-pixel images with the same contrast-to-noise ratio as conventional radiology at that ultra-low photon level. The approach also achieved the maximum contrast-to-noise ratio attainable with direct radiology while using 100 times fewer photons. The authors describe the results as experimental evidence that dose-efficient, high-detail imaging is feasible.

The researchers note that further development is needed before clinical use. Planned work includes improving image quality and demonstrating the approach with laboratory X‑ray sources such as X‑ray tubes to move beyond synchrotron‑based experiments.

“While traditional X-ray imaging relies on enough X-ray photons reaching a detector to form a clear image, our approach uses computational techniques to reconstruct an image from fewer photons. We were able to show the low-dose potential of this approach by achieving megapixel radiology with ultra-low-light,” said Tiqiao Xiao, research team leader at the Shanghai Advanced Research Institute, Chinese Academy of Sciences.

“Our technology could be combined with routine hospital imaging equipment such as chest X-rays and CT scans. It would make medical X-ray imaging safer, which is especially important for children, pregnant patients and people needing frequent scans,” added Xiao.

Related Links
Shanghai Advanced Research Institute, Chinese Academy of Sciences