What we're working on next

Conventional MRI sessions take 15–45 minutes for conventional protocols, and considerably longer for whole-body imaging. We are investigating whether physics-based simulation approaches could meaningfully reduce acquisition time.

In early development testing on phantom data, we have observed that protocols capable of supporting synthetic reconstruction of multiple MRI contrasts can be acquired in approximately three minutes, compared with 30+ minutes for an equivalent conventional protocol.

We are also investigating whether comparable approaches could be extended to whole-body imaging within five minutes.

Most clinical MRI today produces images that are interpreted qualitatively. Two scans of the same patient on different scanners — or even the same scanner at different times — are difficult to compare numerically.

This limits MRI's role in tracking disease progression and treatment response over time.

We are investigating reconstruction approaches that could yield quantitative tissue characterizations as a routine output of clinical imaging, in a form that is robust across scanners and acquisitions.

Radiology workloads are growing faster than the radiology workforce. Computational tools, including machine learning–based ones, are increasingly being explored as workflow aids, ranging from automated reconstruction and quality control to flagging cases or regions for review.

Corsmed is investigating how the structured, physics-grounded outputs of reconstruction-based MRI might serve as better inputs to computational tools that automate routine technical steps and support, but do not replace, radiologist interpretation.