Novel AI Method Sharpens 3D X-ray Vision (www.bnl.gov)

Scientists at the NSLS-II have developed a groundbreaking AI method, the perception fused iterative tomography reconstruction engine (PFITRE), which enhances 3D X-ray imaging capabilities. This innovative technique addresses the "missing wedge" problem, a limitation in traditional X-ray tomography that results in blurry images due to incomplete data collection from all angles. By integrating a convolutional neural network with physics-based models, PFITRE provides clearer reconstructions of nano-scale features in objects like microchips and battery materials, marking a significant leap in imaging resolution—over 10,000 times greater than standard medical CT scans. The implications for the AI/ML community are profound. PFITRE not only improves the accuracy of X-ray imaging but also opens up new avenues for research in various fields, including microelectronics, material science, and biomedical applications. Its ability to analyze complex samples that were previously considered impossible to image could lead to advancements in defect diagnosis and new material synthesis. As AI continues to evolve in conjunction with synchrotron science, techniques like PFITRE will play a crucial role in transforming how scientists explore and understand the microscopic world, potentially addressing significant scientific challenges in the future.