AI Tool for Pancreatic Cancer Gets FDA Nod
An artificial intelligence tool designed for pancreatic cancer risk assessment has earned Breakthrough Device designation from the FDA. This status is intended to expedite the development and review of technologies for life-threatening diseases. The designation reflects the growing clinical and regulatory focus on AI-driven diagnostic tools in oncology.
- The AI tool, named DAMO PANDA, was developed by Alibaba's research institute, DAMO Academy. It demonstrated 92.9% sensitivity and 99.9% specificity in detecting pancreatic ductal adenocarcinoma (PDAC) lesions in a validation study involving over 20,000 patients. The technology is designed to identify subtle signs of cancer on non-contrast CT scans, which could turn routine abdominal scans for other issues into opportunities for early cancer screening. - A key advantage of the DAMO PANDA tool is its effectiveness with non-contrast CT scans. This is significant for outpatient and mobile imaging providers as it exposes patients to lower radiation doses, minimizes risks from contrast agents, and reduces costs compared to contrast-enhanced CT, the current standard for pancreatic evaluation. - The outpatient imaging market is growing faster than the overall radiology market, with advanced imaging volume projected to increase by 14% over the next decade. Technologies like this AI tool align with the trend of shifting care to more convenient and lower-cost community settings, with about 40% of imaging now performed in outpatient centers. - Competitors include PANCREASaver, a tool developed by National Taiwan University Hospital that has also received FDA Breakthrough Device designation and is already in clinical use in Taiwan. PANCREASaver reports an overall diagnostic accuracy of over 90% and is integrated directly into the hospital's PACS system. - For imaging providers, a major operational consideration is workflow integration. AI tools must be integrated into existing IT infrastructure, such as PACS and RIS, without disrupting established reading practices. The lack of standardization for how AI results are integrated and modified by radiologists remains a challenge for widespread adoption. - While FDA clearance for AI tools is increasing, reimbursement remains a hurdle. Most AI applications in radiology have Category III CPT codes, which are temporary codes for tracking usage and are not tied to payment. Gaining dedicated Category I CPT codes, which are necessary for reimbursement, is a separate, lengthy process.
