Radiology Partners Acquires Stanford AI Startup for $80M

Key Points

• Radiology Partners acquired Stanford-founded Cognita Imaging for $80 million
• Cognita’s AI technology cuts reading time by up to 76% while boosting accuracy
• The collaboration gives AI developers access to 55 million annual imaging studies
• Mosaic Drafting tool already tested across 95,000 patient scans

Background

Radiology Partners, America’s largest radiology practice, just made one of the biggest moves in medical AI history. In November 2025, the Nashville-based company acquired Cognita Imaging, a Stanford researcher-founded startup, for $80 million (ℹ️ Radiology Business).

Founded in 2024 by Stanford University scientists and led by CEO Louis Blankemeier, Cognita emerged from “stealth mode” with this announcement. The company spent its early days quietly developing vision language models—a sophisticated type of AI that can analyze medical images almost like a radiologist would.

What Happened

This isn’t just another tech acquisition. Radiology Partners and Stanford collaborate on AI development through this deal, creating a powerful partnership that combines cutting-edge research with real-world medical practice.

Radiology Partners includes more than 4,000 radiologists reading more than 55 million images every year (ℹ️ STAT News). That’s a massive amount of data—and it’s exactly what Cognita needs to train and improve its AI models through FDA trials.

The technology behind this deal is called Mosaic Drafting, an AI tool that analyzes X-rays and head CT scans, then drafts preliminary radiology reports. Think of it as a smart assistant that does the first pass, letting radiologists review, edit, and approve the final results.

The numbers are impressive. According to Radiology Partners, the technology has demonstrated up to 52% increased detection rates, a fourfold reduction in significant diagnostic errors, and up to 76% shorter read times (ℹ️ AuntMinnie). Even better, radiologists actually like using it—between 87% and 98% of AI-generated results needed just one or fewer clinically significant edits.

Why It Matters

This acquisition matters for several reasons. First, it shows that AI isn’t replacing radiologists—it’s helping them work smarter and faster. Radiologists still maintain full oversight and make all final decisions.

Second, the transaction creates what business experts call a “data flywheel.” More data helps train better AI, which attracts more partners, generating even more data. The deal gives Cognita access to Radiology Partners’ dataset of 55 million annual imaging studies as it advances its models through FDA trials (ℹ️ Radiology Business).

For patients, the result could mean faster diagnoses and fewer missed findings. For healthcare systems facing radiologist shortages, it could help existing doctors handle more cases without burning out.

The technology uses vision language models, which are more versatile than traditional AI. Instead of looking for one specific problem, these models can scan for any finding that looks important—similar to how a human radiologist thinks.

What’s Next

Cognita will continue operating as an independent business unit within Radiology Partners, maintaining its AI focus while integrating with the larger company’s MosaicOS operating system.

Cognita’s models are currently undergoing Institutional Review Board-approved trials as part of the FDA authorization process (ℹ️ AuntMinnie). Once approved, the technology could roll out more widely across hospitals and imaging centers.

Radiology Partners, backed by Whistler Capital Partners and New Enterprise Associates, employs over 4,000 radiologists serving more than 3,400 hospitals and healthcare facilities. The company officially launched its Mosaic technology services division in July 2025.

This acquisition represents a major step forward in medical AI, bringing together Stanford’s research expertise with Radiology Partners’ massive real-world testing ground. As AI continues evolving in healthcare, partnerships like this one show how technology can enhance—not replace—human medical expertise.

Source: Multiple authoritative sources including STAT News, Radiology Business, and AuntMinnie — Published November 2025
Original sources:

• STAT News
• AuntMinnie