FDA Approves First AI-Enabled Imaging Device for Breast Cancer Surgery Margin Assessment
Key Insights
The FDA has granted premarket approval to Claire (search), the first AI-enabled imaging device in the United States for intraoperative breast cancer (search) margin assessment during breast-conserving surgery.
Claire (search) demonstrated 88.1% margin accuracy and statistically significant reduction in patients with residual cancer post-surgery compared to standard of care alone in a pivotal trial of 206 patients.
The device delivers 10 times higher resolution than X-ray or ultrasound at 2mm imaging depth and was trained on over 2 million breast tissue images from Perimeter's proprietary OCT library.
The FDA has granted premarket approval to Claire (search) (formerly known as Perimeter OCT B-Series with ImgAssist AI 2.0), marking a significant milestone as the first AI-enabled imaging device approved in the United States for intraoperative breast cancer (search) margin assessment. The technology, developed by Perimeter Medical Imaging AI (search), Inc., is designed to enhance surgeons' ability to detect difficult-to-see cancer during breast-conserving surgery for patients with stage 0 to III invasive ductal carcinoma (search) and/or ductal carcinoma in situ (search).
Breakthrough Technology Addresses Critical Clinical Need
Claire (search) combines Perimeter's proprietary AI with patented wide-field OCT imaging to enable high-resolution, real-time evaluation of excised tumor margins. The system delivers 10 times higher resolution than standard X-ray and ultrasound at 2mm imaging depth - the clinically relevant margin width for breast cancer (search) margin assessment. The AI technology was trained on Perimeter's proprietary and growing OCT image library of over 2 million breast tissue images.
"Despite progress in breast cancer (search) treatment, intra-operative margin assessment remains challenging, often leading to excess removal of healthy tissue, re-operations, and anxiety for patients as they await pathology results," said Dr. Alastair Thompson, Surgeon and Professor, Section Chief of Breast Surgery at Baylor College of Medicine and Primary Principal Investigator of the pivotal trial. "Claire (search) has the potential to become a new standard in breast surgical care helping reduce re-excisions while improving patient outcomes."
Currently, national averages indicate repeat surgeries occur in about 20 percent of breast-conserving surgeries in the United States. Today, surgeons primarily rely on physical examination and limited intraoperative pathology to assess margins, leaving patients to wait up to a week or more to learn whether margins are clear or additional surgery is required.
Pivotal Trial Demonstrates Superior Performance
The pivotal trial results, presented at the 26th Annual Meeting of the American Society of Breast Surgeons, demonstrated Claire (search)'s clinical efficacy across 206 evaluated patients. Of these patients, 56 had residual disease margins in 17% of patients after intraoperative standard of care alone, which consisted of a lumpectomy.
In 40% of patients, the use of Claire (search) after standard of care evaluation resulted in the correct detection of residual disease, and fully clearing 20% of additional patients from all residual disease. This met the pre-specified super-superiority performance goal for the primary endpoint (P = .0050). The overall Claire accuracy was 88.1%.
In the trial arm utilizing Claire (search), the mean total lumpectomy volume excised was 74 cm³, including 56.5 cm³ from primary lumpectomies, 14.7 cm³ from standard of care shaves, and 2.8 cm³ from 115 Claire shaves. A total of 26 patients showed a clinical benefit rate from the identification of residual disease by Claire after standard of care was completed.
Notably, 6 patients who had Claire (search)-aided shaves contained pathology-confirmed disease that was missed by standard of care and histopathology at the previous margin, highlighting the technology's ability to detect cancer that conventional methods might overlook.
AI Innovation and Regulatory Framework
A critical innovation behind Claire (search) is its AI engine, which represents one of the few AI-enabled class III devices on the U.S. market today. The system's unique advantage lies in its ability to continuously improve, as every surgical procedure performed with Claire generates new data that can be used to enhance the AI product for better outcomes in future patients.
The FDA approval includes a pre-determined change control plan that allows for AI enhancements without the need for additional FDA authorization or interaction. This framework enables the system to evolve and improve as Perimeter's database expands with more surgical procedures.
"Developing AI for clinical use is incredibly challenging, requiring large, high-quality datasets that reflect real-life patient diversity," said Andrew Berkeley, Perimeter's Chief Innovation Officer and co-founder. "This approval reflects years of hard work with our clinical partners integrating AI directly into surgical decision-making and advancing its role from operational support to real-time patient care."
Commercial Launch and Market Impact
Perimeter plans to begin a nationwide launch in the coming weeks, targeting an estimated 300,000 annual U.S. breast cancer (search) surgeries. The company's patented wide-field OCT technology, proprietary dataset, and integrated AI capabilities provide a scalable foundation for potential expansion into additional cancer indications over time.
"Repeat breast cancer (search) surgeries due to residual disease remain a significant clinical, health, and economic burden," said Adrian Mendes, Perimeter CEO. "Claire (search)'s FDA approval marks a major milestone in breast cancer care, as we advance our goal of reducing repeat surgeries so that no patient has to be told 'we didn't get it all.'"
The pivotal trial was supported by a US$7.4 million grant from the Cancer Prevention and Research Institute of Texas (search) (CPRIT), enabling the large-scale clinical validation necessary for FDA approval.