We started by listening. Radiologists told us that crossview correlation in DBT is a definite challenge, so we built a fourth generation, proprietary AI expressly for the human-in-the-loop moment: precise tissue matching that feels natural and fast. We’re fortunate that in this case, clinical and financial benefits overlap: evidence for better localisation and confidence on one side, and a clear reimbursement potential on the other. That alignment is rare, and it’s why we’re moving quickly to bring ViewFinder to everyday practice.
Stephen Morrell, Founder & CEO, Elaitra, Visiting Research Fellow, School of Biomedical Engineering & Imaging Sciences
26 November 2025
ViewFinder AI brings cross view precision to 3D mammography
Elaitra, a medical imaging startup founded by an international collaboration including King’s College London researchers Stephen Morrell and Jorge Cardoso, has launched ViewFinder, a new FDA-cleared AI tool designed to make breast cancer screening using digital breast tomosynthesis (DBT) easier and quicker for radiographers.
DBT screening is an advanced form of mammography that produces 3D images. It offers a variety of benefits including improved accuracy and sensitivity over conventional 2D scanning, which is currently used as part of the NHS national breast screening programme. Accurately matching the same tissue area across different views however, remains a persistent issue for radiologists interpreting DBT results.
ViewFinder helps to overcome this challenge by providing real-time visual guidance as radiologists interpret images. When a radiologist hovers over an area, ViewFinder identifies the corresponding location in other views and displays a confidence-based outline as well as a front-to-back ‘coronal’ visualisation. This dynamic assistance improves 3D navigation and helps to reduce guesswork, particularly for very small lesions where precision is critical.
This marks one of the first breast imaging solutions to embody the next wave of ‘interactive’ or ‘embodied’ AI – systems that work alongside clinicians in real time instead of running silently in the background. By enhancing spatial reasoning, the tool improves diagnostic accuracy and communication during work-ups.
Everyone is talking about ‘embodied AI’. ViewFinder is one of the first breast imaging AIs to use embodied AI well. It augments the radiologist’s spatial reasoning rather than trying to replace it. By focusing on human–AI interaction, the team has cracked a major problem in breast cancer imaging with a new, proprietary approach.
Dr Jorge Cardoso, Reader in Medical Imaging, School of Biomedical Engineering & Imaging Sciences, Chief AI Officer of Elaitra
A peer-reviewed study published in the European Journal of Radiology (EJR) shows that ViewFinder significantly reduces localisation errors across normal, benign and abnormal cases. The largest improvements were seen among less experienced readers, while radiologists of all skill levels reported higher confidence. Because the technology is backed by published evidence, it also qualifies for immediate ‘by report’ reimbursement in the USA – aligning the technology’s clinical benefits to commercial value.
Elaitra is also recruiting radiologists for an international follow-up study to further expand evidence across diverse practice environments and for reimbursement pilots in the USA.
“ViewFinder is a great example of our bench to bedside to boardroom approach. By using AI to help break down barriers for radiologists using 3D mammography imaging Elaitra has developed a commercially viable product that makes using latest technology much easier for clinicians and improves the accuracy of scans and diagnoses for patients.”
Professor Sebastien Ourselin FREng FMedSci, Assistant Principal (Innovation), Head of School of Biomedical Engineering & Imaging Sciences