The Food and Drug Administration has cleared a new artificial intelligence tool designed to help doctors find a notoriously hard-to-spot cause of gastrointestinal bleeding: lesions hiding deep in the small bowel, a stretch of intestine too long and too remote for a standard endoscope to reach. The clearance, granted to California-based CapsoVision on September 25, 2026, covers a feature called AI Highlights built into the company’s CapsoCam Plus capsule endoscopy system.
A Pill-Sized Camera Meets a Pattern-Matching Algorithm
CapsoCam Plus is a disposable, pill-sized capsule that a patient swallows. As it travels through the digestive tract, it records continuous video of the small bowel, a region that sits beyond the reach of traditional upper endoscopy and colonoscopy and has historically been one of the most difficult parts of the gut to evaluate. The capsule generates tens of thousands of images per study, and until now, a gastroenterologist had to scroll through nearly all of them to find the handful of frames that mattered.
AI Highlights changes that workflow. During video processing, the algorithm scans every frame captured by the capsule and flags images containing suspected abnormal lesions, prioritizing them for physician review rather than requiring a frame-by-frame manual search. According to CapsoVision, the tool is cleared for use in adult and pediatric patients older than two years whose capsule study was ordered because of suspected small bowel bleeding, one of the most common reasons physicians order the test.
Why Small Bowel Bleeding Is So Hard to Diagnose
Obscure gastrointestinal bleeding, meaning bleeding whose source can’t be identified by conventional endoscopy or colonoscopy, accounts for a meaningful share of GI referrals, and a large portion of those cases trace back to lesions in the small bowel. Capsule endoscopy was developed specifically to address that blind spot, but the sheer volume of images each study produces has made manual review slow and dependent on the reviewing physician’s experience and attention span. Subtle angioectasias or small ulcers can be easy to miss in a sea of otherwise unremarkable frames.
CapsoVision has pitched AI Highlights as a way to compress review time without asking physicians to cede diagnostic authority to software. The company’s clearance documentation and public statements are explicit that the tool is intended to assist, not replace, a gastroenterologist’s interpretation, and it is not meant to substitute for histopathological sampling when tissue diagnosis is needed. In practice, that means the AI narrows down which frames a physician looks at closely, rather than issuing its own diagnosis.
A Global Rollout, With the U.S. Arriving Last
The U.S. clearance is notable partly because of its timing. CapsoVision had already launched AI Highlights internationally, including across the European Union under the bloc’s Medical Device Regulation, before pursuing FDA clearance domestically. That sequencing is common among device makers navigating the different regulatory tracks in the U.S. and Europe, but it also means U.S. gastroenterologists are getting access to a tool that clinicians elsewhere have already been using in practice, giving CapsoVision a body of real-world use to point to as it markets the feature domestically.
The clearance arrives alongside broader updates to CapsoVision’s CapsoView and CapsoCloud software, suggesting the company is treating AI Highlights as one piece of a larger push to modernize how capsule endoscopy data gets stored, shared, and reviewed across hospital systems and outpatient GI practices.
The Workflow Argument, and Its Limits
Supporters of AI-assisted capsule review argue that the technology addresses a real bottleneck: physician time. A single small bowel capsule study can take a reviewer the better part of an hour to examine thoroughly, and backlogs in GI practices mean results can take days to reach a referring physician. Tools that triage frames for review have the potential to shrink that turnaround time substantially, which matters for patients who are actively bleeding or anemic while waiting for an answer.
But the same caution that has accompanied other FDA-cleared diagnostic AI tools applies here. Any triage algorithm carries a risk of false negatives, images with real pathology that the software doesn’t flag and that a time-pressed reviewer might skip past if they lean too heavily on the AI’s prioritization. Gastroenterology societies have generally urged that AI-assisted review supplement, not shortcut, full study review, especially in early years of adoption as real-world performance data accumulates outside of the controlled validation studies submitted to the FDA.
What Comes Next
CapsoVision’s clearance adds to a small but growing list of FDA-authorized AI tools aimed specifically at gastrointestinal diagnostics, following earlier clearances for AI-assisted polyp detection during colonoscopy from other manufacturers. As more capsule and endoscopy makers build AI into their core software rather than selling it as an add-on, the expectation among industry watchers is that frame-triage tools like AI Highlights will become a standard feature rather than a premium one. For hospitals and GI practices still relying on fully manual capsule review, the clearance is likely to intensify pressure to upgrade, particularly as insurers and health systems look for ways to speed up diagnostic turnaround without adding headcount.
Photo: Aviavlad / PIXABAY via Pixabay
