Study Analyzes Real-Time Detection of Solid Pancreatic Masses Using AI-Enhanced Endoscopic Ultrasonography

Study Analyzes Real-Time Detection of Solid Pancreatic Masses Using AI-Enhanced Endoscopic Ultrasonography

October 26, 2026

Orlando Health Digestive Health Institute, in partnership with Medical Softverse in Romania, has developed the world’s first artificial intelligence integrated endoscopic ultrasound (AI-EUS) program designed to detect both solid and cystic pancreatic lesions.

Shyam S. Varadarajulu, MD

Called PANCRAIEUS, the imaging analysis software was tested in a single-center prospective, non-randomized, comparative study at the Orlando Health Center for Advanced Endoscopy, Research and Education (CARE). Findings were published in the September 2025 issue of Endoscopy, the premier journal covering gastrointestinal endoscopy.

“Endoscopic ultrasonography (EUS) is the most sensitive modality for accurately establishing a tissue diagnosis in patients with solid pancreatic masses,” says Shyam S. Varadarajulu, MD, president of Orlando Health Digestive Health Institute. “While this technology is critical for imaging, staging and diagnosing pancreatic cancer, detection of small lesions can be challenging, particularly for less experienced endosonographers. The AI-EUS program was designed to optimize procedural performance to better standardize the early detection of solid mass lesions and improve outcomes for pancreatic ductal adenocarcinoma.”

Patients at risk for pancreatic cancer or with suspected pancreatic lesions were evaluated in real-time using the AI-enhanced EUS software created in collaboration with a team directed by Professor Adrian Saftoiu in Bucharest, Romania. Among 308 patients referred for evaluation from January through July 2024, AI-EUS achieved a detection rate of 97.1% for pancreatic mass lesions vs 100% for conventional EUS performed by experts.

“Our initial study results concluded AI-EUS performance did not differ significantly from conventional EUS performed by experienced endosonographers,” says Dr. Varadarajulu. “However, it demonstrated that adding AI to EUS could potentially decrease the learning curve and improve technical performance of novice endoscopists, thereby optimizing clinical outcomes in pancreatic cancer.”

Undertaken at a single center with only expert endosonographers, the research will require external clinical validation, notes Dr. Varadarajulu. Plans for a multicenter study at other tertiary-care academic centers are currently in development to further refine the new technology.

A discussion of this study featuring the trial investigators is available here.