Quantum hybrid method cuts MRI scan time by 80% using D-Wave
D-Wave quantum hybrid matches full-quality MRI at just 10% sampling rate
An interdisciplinary team including Asmit Ganguly and Danny J.J. Wang has published a new method on arXiv that applies quantum adaptive sensing to accelerate MRI. The framework reformulates the problem of selecting Cartesian phase-encode lines as a quadratic unconstrained binary optimization (QUBO) problem, which can be solved using classical annealing or quantum-annealing hardware like D-Wave systems. The objective function balances central k-space preference, signal-energy information from prior measurements, and pairwise dispersion terms to avoid redundant sampling.
In retrospective experiments using simulated eight-coil 3D MRI data at 20% and 10% sampling rates, the QUBO-based adaptive approach consistently outperformed static variable-density Poisson-disc sampling across metrics including PSNR, SSIM, NMSE, and HFEN. A reduced-pool study using a D-Wave quantum-classical hybrid solver achieved reconstruction quality comparable to static Poisson-disc sampling, confirming feasibility on current quantum infrastructure. The authors note that no quantum computational advantage was observed, but the direct QUBO representation provides a practical framework that could benefit from future quantum-annealing hardware improvements. Prospective scanner validation and systematic benchmarking remain necessary before clinical translation.
- Uses quadratic unconstrained binary optimization (QUBO) for adaptive k-space line selection in MRI.
- Improved image quality metrics (PSNR, SSIM) at 20% and 10% sampling versus state-of-the-art static methods.
- D-Wave quantum-classical hybrid solver achieved comparable quality to variable-density Poisson-disc, demonstrating feasibility on current hardware.
Why It Matters
Enables faster, cheaper MRI scans by leveraging quantum annealing for optimal sampling, potentially reducing patient discomfort and costs.