NSF AI grant $1.25M: AI for low-dose, low-cost 3D X-ray (CT) — SBIR Phase II (surgical guidance)

The NSF awarded about $1,250,000 to a Small Business Innovation Research (SBIR) Phase II project for AI-powered low-dose, low-cost, high-quality computed tomography (CT) imaging (NSF Award 2604163; program: SBIR Phase II; California; starting June 2026).

Per the abstract (broader impact/commercial potential), the goal is to expand access to advanced 3D X-ray imaging during surgery without requiring hospitals to purchase expensive CT scanners. Many procedures, particularly spine operations, rely on 2D X-ray images that can make it hard to fully visualize anatomy and implanted hardware; limited access to affordable 3D imaging can increase procedure time, complication rates, and costs. The project enables existing mobile X-ray systems to produce high-quality 3D images, allowing more procedures to be safely performed in outpatient surgical centers and community hospitals. If successful, it could reduce healthcare expenditures, improve patient safety, and lower radiation exposure by avoiding repeat scans. Commercially, it supports a scalable, software-based model that upgrades widely deployed imaging equipment rather than replacing it.

Technically, the project develops and clinically validates a method to generate 3D images from limited-angle X-ray data acquired by standard mobile surgical imaging systems. Because conventional mobile systems rotate over a small angle and operate under dose constraints, they primarily produce flat 2D images and struggle with accurate 3D reconstruction. The project refines AI models that incorporate physical principles of X-ray imaging to reconstruct volumetric images from limited data. Objectives include improving image quality and reliability across different systems and patient anatomies, developing real-time calibration to correct for mechanical motion and geometric distortion, and validating performance in realistic surgical environments. The anticipated results are rapid, low-dose 3D reconstructions with clarity and geometric accuracy comparable to conventional CT for specific surgical tasks — a practical path to advanced 3D guidance using equipment already widely available in operating rooms.