A research team from NYU Tandon has developed a breakthrough algorithm that could dramatically accelerate stroke detection by making portable, real-time brain imaging feasible for the first time. Today, determining whether a stroke is caused by a clot or a bleed requires large CT scanners, machines that are unavailable in many ambulances, rural hospitals, and low-resource settings. As a result, countless patients lose precious treatment time.
Microwave imaging has long been viewed as a promising alternative. The technology can detect changes in the brain’s electrical properties without radiation or heavy equipment, offering a path toward compact, helmet-sized scanners. But one challenge has stalled clinical use: image reconstruction has simply been too slow. “Portable hardware wasn’t the problem, the computation was,” says Stephen Kim, Research Professor at NYU Tandon. “Waiting nearly an hour for a result is not an option during a stroke emergency.”
10 to 30 times faster
In a study published in IEEE Transactions on Computational Imaging, Kim, Ph.D. student Lara Pinar, and department chair Andreas Hielscher introduce an advanced reconstruction algorithm that is 10 to 30 times faster than existing methods. By rethinking the underlying mathematics, the team eliminated the need for laborious intermediate calculations. Instead, the algorithm starts with rapid, rough approximations and refines them only when necessary. An approach that dramatically reduces computational load.
Additional optimizations, including compressed mathematical representations and more stable modeling for complex head shapes, further boost performance. Reconstructions that once took an hour now appear in under 40 seconds, while maintaining high image quality across experimental datasets. According to Hielscher, this marks a turning point. “Microwave imaging has always had the potential to be portable and affordable. What we lacked was speed. Now we’re finally bridging that gap and moving toward real clinical use.”
The implications go far beyond stroke diagnostics. Rapid, portable microwave scanners could:
- offer an accessible breast cancer screening alternative where mammography is unavailable,
- monitor brain swelling in the ICU without repeated CT scans,
- assess tumor response in oncology by detecting subtle tissue changes.
The NYU team is already working on extending the algorithm to full 3D imaging. This is an essential step toward commercial deployment and bedside use. “We’re giving a powerful technology the speed it needs to matter in real-world care,” Kim says. “And that opens the door to helping patients who currently have no access to timely diagnostic imaging.”
Speeding up stroke care
Earlier this year we reported on a simulation model, created bij researcher at Erasmus MC, to evaluate whether a mobile stroke unit (MSU) could improve care for stroke patients in the Rotterdam region. Their findings showed that an MSU would enable suspected stroke patients to receive diagnosis and treatment significantly faster than under current protocols.
Speed is critical in stroke care, as brain cells die within minutes when blood flow is disrupted. Today, patients must first reach a hospital for a CT scan to determine whether the stroke is caused by a clot or a hemorrhage, an essential distinction that delays treatment.
An MSU functions as an ambulance equipped with a CT scanner, specialized tools, and a trained stroke nurse. When a suspected stroke is reported, both a standard ambulance and the MSU respond. The on-board nurse, together with a remote neurologist, can immediately perform diagnostics and begin treatment, then transport the patient to the most appropriate hospital.
