GE HealthCare Receives FDA Authorization for Deep-Learning Mammography Reconstruction Tool

0
2

CHICAGO — GE HealthCare has received FDA Premarket Authorization for Pristina Recon DL, a new deep-learning–based image reconstruction technology designed to improve the clarity and detail of 3D mammography scans. The company says the system produces sharper images with fewer artifacts, offering radiologists a more consistent view during breast cancer screening and diagnosis.

Pristina Recon DL is built into GE’s Pristina Via mammography platform and uses two sequential deep-learning models. The first reconstructs high-fidelity 3D volumes and reduces noise, while the second enhances clinically relevant features in the synthesized 2D images produced from the tomosynthesis scan. According to GE, the system is the first to combine deep learning with iterative reconstruction for breast imaging without raising radiation dose.

Breast cancer remains one of the most common cancers among women, with one in eight expected to receive a diagnosis during their lifetime. Global deaths are projected to reach more than one million annually by 2050, underscoring the need for improved tools for earlier detection and greater diagnostic accuracy.

In a recent reader study cited by GE HealthCare, breast radiologists preferred images reconstructed with Pristina Recon DL in 99 percent of comparisons with the previous generation of reconstruction technology. The company also said the system demonstrated better performance in detecting microcalcification clusters and masses in modeled clinical testing.

The new reconstruction tool runs on NVIDIA RTX accelerated computing hardware, enabling rapid processing intended to support real-time clinical workflows. GE HealthCare says the system also maintains the Pristina Via platform’s emphasis on efficient exams, including personalized protocols, zero-click image acquisition and no added wait time during reconstruction.

Executives and clinical partners said the technology addresses longstanding challenges in digital breast tomosynthesis. “By applying advanced deep learning technologies, we’re shaping the future of breast imaging – one defined by uncompromised image quality, faster workflows, and greater confidence in early cancer detection,” said Jyoti Gupta, PhD, President and CEO of Women’s Health and X-ray at GE HealthCare.

Howard Berger, MD, President and CEO of RadNet, which collaborated with GE on evaluation of the technology, said the improvement in 3D image quality represents “a meaningful upgrade” for radiologists and patients.

The company said the broader Pristina Via system maintains high uptime, accommodates a wide range of patient needs and supports additional clinical applications, including contrast-enhanced mammography and breast biopsy workflows.

GE HealthCare worked with academic institutions and high-volume imaging centers throughout the development and validation of the algorithms. Clinical groups that tested the system said the upgraded imaging clarity strengthens diagnostic confidence and improves overall breast cancer care.

Leave A Reply

Please enter your comment!
Please enter your name here