Reading a head and neck CT angiogram usually means waiting on a separate post-processing team to strip bone and veins, and that step now has an FDA-cleared shortcut. Deep Reasoning AI, founded by Cornell University artificial intelligence professors, won 510(k) clearance on October 27 for NeuroDR, software that renders the arteries of the head and neck from CTA scans.
The clearance, number K262134, covers a tool running in the background the moment a scan finishes. What returns to the hospital picture archive as DICOM files are curved reformats, volume renderings and 3D maximum intensity projections of the internal carotid and vertebral arteries on both sides. That means separating bone, soft tissue and veins first. Front and back flow stay in separate views, so neither hides anatomy within the other.
The cleared role is assistive. NeuroDR helps a doctor confirm whether a lesion they already spotted is present or absent, and supports follow-up plans and documentation. It does not hunt for lesions alone, which keeps the product in imaging AI’s assistive tier.
The imaging work was joint. Performance studies came from 3DR Labs, which calls itself the biggest 3D medical image post-processing provider in the United States, and NeuroDR was built with it. Innolitics, a medical device software and regulatory consultancy, shaped those studies and drafted the filing and cybersecurity paperwork. It knew what the agency expected from similar cleared devices, so NeuroDR skipped the pre-submission meeting and got a decision 105 days after filing.
Deep Reasoning AI chief executive Mike Beck called it the company’s first cleared product and a step toward full patient-specific 3D anatomy from routine scans. Innolitics partner Yujan Shrestha, MD, credited a client ready to do the work FDA asks for. Both firms plan to bring it to U.S. hospitals and imaging centers.
