GE HealthCare Receives FDA 510(k) Clearance for Next-Gen MRI Imaging Technology AIR Recon DL with Expanded 3D and PROPELLER Capabilities

Source: Instrument Family, no reprinting without authorization, and reprinting is allowed 24 hours later.

Recently,器械之家 learned that GE Healthcare announced the U.S. FDA 510(k) clearance of its groundbreaking AIR Recon DL for 3D and PROPELLER imaging sequences.These new features extend the advantages of AIR Recon DL to almost all magnetic resonance imaging (MRI) clinical procedures, covering all anatomical structures, achieving better image quality, shorter scan times, and an improved patient experience.

GE Healthcare's AIR Recon DL, an MRI image reconstruction technology based on deep learning, has fundamentally transformed the balance between image quality and scan time.GE Healthcare said in a press release that now, with the extended compatibility of this solution from 2D to 3D imaging sequences, doctors can diagnose patients quickly and confidently, improving the signal-to-noise ratio (SNR) and clarity. 3D imaging provides higher clinical efficiency, eliminating the need for radiologists to perform multiple 2D acquisitions, which could potentially accelerate the diagnostic process.

01

What kind of technology is AIR Recon DL?

AIR™ Recon DL is the deep learning reconstruction technology pioneered by GE Healthcare and the first in the industry to receive FDA approval.In AIR™ Recon DL, after tens of millions of fitting iterations, the deeply optimized deep convolutional neural network model overturns traditional reconstruction concepts. This technology is the first to embed deep learning into the raw data reconstruction phase of magnetic resonance imaging., the effective separation of MR signals from noise signals is achieved, thereby obtaining pure magnetic resonance signals, significantly improving the efficiency of magnetic resonance imaging, making the selection of scanning parameters no longer a major factor limiting image quality, and broadening new boundaries for magnetic resonance scanning speed, resolution, and signal-to-noise ratio.

AIR Recon DL was co-developed with global institutions and has been evaluated in thousands of cases across various anatomical structures and patient demographic data. Feedback from clinical users has been highly positive, with users noting clearer images with less noise, leading to reduced scan times, increased confidence, decreased need for repeat scans, and improved consistency between scans.

The development and clinical validation partners of AIR Recon DL include:Hospital for Special Surgery, University of California, San Francisco, RadNet, University of Wisconsin-Madison, MD Anderson Cancer Center, Medical College of Wisconsin, Centre Cardiologique du Nord, Erasmus Medical Center, Centro Cardiologico Monzino, University of Yamanashi and Keio University, a San Medical Center and Haeundae Paik Hospital.

AIR Recon DL was developed on GE Healthcare's Edison Intelligence Platform, which enables GE and strategic partners to quickly design, develop, manage, secure, and distribute advanced applications and AI algorithms.

AIR™ Recon DL AI Platform Was First Introduced by GE Healthcare to the MRI Reconstruction Process in 2019 and Received FDA Approval for Clinical Use on 3.0T Systems.In AIR™ Recon DL, the defined multi-layer deep convolutional neural network undergoes tens of millions of fitting iterations to obtain a set containing millions of weighting factors. Based on the optimized deep convolutional neural network model, the data acquired by magnetic resonance imaging differs from traditional reconstruction methods.Its valid information will be automatically extracted for reconstruction. The obtained images have the following characteristics:

1. Significantly reduced noise levels, thereby improving image signal-to-noise ratio. Note: The choice of noise reduction level can be based on the user's specific preferences, including three options: 25%, 50%, and 75%.

2. Significantly improved image resolution and sharper inter-tissue contrast;

3. Significant reduction in image artifacts caused by motion and other reasons.

The increased image signal-to-noise ratio and enhanced image contrast make the setting of scanning parameters less demanding during the scanning process, allowing the scanning speed to be increased by 30%-50%., thereby better serving clinical needs. Combining these advantages, the technology has been deeply explored in structural and functional imaging of various parts of the body, including the nervous system, breast, torso, bones, and muscles, yielding fruitful scientific research results.

02

Taking the nervous system as an example

Solve the Last Kilometer Problem in Clinical Practice

Seconds Replace Minutes as the New Unit of Measurement for Research Scanning; Hundred-Second Imaging Typically Encompasses Sub-Millimeter Voxel Structure Imaging, High-Resolution Functional Imaging, and Quantitative Mapping. Neurological system imaging is the most focused direction in magnetic resonance imaging research.With the underlying reconstruction of magnetic resonance imaging by AIR™ Recon DL, a new balance among scan time, image resolution, and image signal-to-noise ratio has been established, also opening a new chapter for neuroimaging.

Ultra-fast Imaging

Ultra-fast imaging has always been an important direction in the field of magnetic resonance technology.In clinical diagnostic scans, rapid imaging means higher diagnostic efficiency, providing timely imaging evidence for urgent diagnoses, thereby meeting the increasing demand for clinical diagnostics. However, a mere increase in imaging speed often leads to a decline in imaging quality, including reduced signal-to-noise ratio, decreased contrast, and increased image artifacts. This presents a challenge for neurologic disease diagnosis, which particularly relies on high-definition imaging as a diagnostic basis. Therefore, in neuro magnetic resonance imaging (MRI) scans, can rapid and high-quality imaging be achieved simultaneously?

AIR™ Recon DL Technology Provides the Answer:Based on the three major features of AIR™ Recon DL: high signal-to-noise ratio, high image resolution and sharp contrast, as well as more realistic image restoration after artifact removal, the strict scanning parameter settings required for high-quality imaging in routine clinical head MRI sequences are no longer demanding.By applying AIR™ Recon DL reconstruction technology, clinical head MRI protocols require less than 1/4 of the original scanning time (traditional 10-minute imaging can be completed in just 2 minutes with AIR™ Recon DL) to achieve imaging quality comparable to traditional reconstruction images, truly enabling ultra-fast and high-quality imaging.

High-Definition Tissue Imaging

Brain structural imaging with high signal-to-noise ratio and strong tissue contrast is an essential radiological foundation for studying neurodegenerative diseases, epilepsy, multiple sclerosis, and other neurological disorders.T2w-PROPELLER imaging, due to its effective suppression of artifacts caused by human body motion and fluid flow, has been widely recognized in the clinical diagnosis of intracranial nerves. However, the long scan time required for high-resolution T2w-PROPELLER imaging to achieve sufficient image signal-to-noise ratio and tissue contrast poses challenges for its widespread clinical application.

Combined with AIR™ Recon DL reconstruction technology, the image resolution reaches ultra-high-definition T2w-PROPELLER at the sub-millimeter level (0.2 x 0.2 x 1mm³, PROPELLER-DL) for brain imaging. Compared to images obtained through traditional reconstruction methods (PROPELLER), there is a significant improvement in image signal-to-noise ratio. More importantly, the microstructures of critical brain functional regions such as the amygdala, hippocampus, and cerebellar areas (deep nuclei, cerebellar cortex, and vermis) are displayed more clearly on PROPELLER-DL.

Neurovascular Imaging

Nerve and vascular tissues are core components of the structural organization of the human brain, and their morphological and functional assessment serves as a crucial basis for diagnosing neurological diseases in clinical practice.Due to their tiny structures, the resolution of nerve and vascular magnetic resonance imaging often needs to reach the order of millimeters or even sub-millimeters.Therefore, the longer scan time required leads to motion artifacts, lower image signal-to-noise ratio, and insufficient tissue contrast, which limit its widespread clinical application.

Combined with AIR™ Recon DL reconstruction technology, 2D DIXON and 3D CUBE-STIR imaging can even reduce scan time.The imaging of brachial plexus nerve roots (DIXON-DL, CUBE-STIR-DL) has shown significant improvements in image signal-to-noise ratio, contrast between nerve roots and surrounding tissues, and the display of local nerve details (including nerve roots, nerve trunks, and nerve bifurcations), providing a solid theoretical basis for its application in neural imaging.

Small Vessel Imaging

Meanwhile, in small vessel imaging, T2* or SWI imaging based on 3D EPI can achieve ultra-high isotropic vascular imaging resolution (0.5mm³) within a scan time of less than 4 minutes. However, the relatively low image signal-to-noise ratio presents certain challenges in the detection and diagnosis of small vessel-related diseases (e.g., microbleeds, calcifications, or iron depositions).

However, with the help of AIR™ Recon DL technology, the reconstructed T2* (T2*-DL) and SWI (SWI-DL) vascular imaging show significant improvements in image signal-to-noise ratio, overall image quality, image contrast, and the depiction of small vessels compared to T2* and SWI imaging obtained through traditional reconstruction methods.Besides, T2*-DL and SWI-DL showed higher consistency in the display of image details after two repeated scan tests.

Precision Quantitative Mapping

MAGiC has been widely recognized for its value in clinical neuroscience research due to its one-stop unlimited contrast structural imaging and comprehensive relaxation quantitative mapping.Further combined with AIR™ Recon DL technology, the raw data used for image synthesis shows significant improvements in signal-to-noise ratio, contrast, and the authenticity of data after artifact removal.

Fine Binding Function Evaluation

Diffusion Magnetic Resonance Imaging, due to its ability to accurately reflect the diffusion properties of tissue molecules, brain network connectivity, and microstructures, is widely used in clinical diagnosis, especially in the field of clinical neurology. However, due to limitations in imaging principles, the acquired diffusion images often cannot simultaneously possess characteristics such as high signal-to-noise ratio, high tissue contrast, minimal imaging artifacts, minor image distortion, and short scanning time.

Among them, PROPELLER Diffusion Imaging (DW-PROPELLER) is characterized by fewer artifacts and less deformation, making it suitable for evaluating tissues with relatively inhomogeneous local magnetic fields, such as the skull base and head and neck.However, obtaining imaging with sufficient signal-to-noise ratio and tissue contrast within a shorter scan time is a bottleneck that hinders its further widespread clinical application.

Without increasing imaging time, the signal-to-noise ratio and tissue contrast of the reconstructed PROPELLER Diffusion (DW-PROPELLER-DL) using AIR™ Recon DL technology are significantly improved compared to traditional reconstruction images, reaching a level comparable to Multi-Shot Echo-Planar Imaging (MUSE). Moreover, it outperforms MUSE imaging in artifact suppression and image distortion control.More notably, the quantitative parameter ADC derived from DW-PROPELLER-DL exhibits smaller value fluctuations and more authentically reflects its molecular diffusion characteristics in model experiments.

03

Enhance the MRI Experience

Can be used for upgrading the vast majority of 1.5T and 3.0T models.

AIR Recon DL is now compatible with PROPELLER, a motion-insensitive imaging sequence that is particularly important for anatomical structures susceptible to motion (such as breathing) during MRI examinations, as well as for pediatric, neurodegenerative, elderly, and claustrophobic patients who have difficulty remaining still during MRI scans.Therefore, doctors can obtain clearer images without repeated scans, thereby increasing patient throughput, improving scheduling, shortening examination times, and accelerating diagnosis, helping to create a better overall patient experience.

The original intention of developing AIR™ Recon DL technology was to tackle the most challenging pain points in clinical practice and solve the last-mile problem in the clinic. As of September 2022, at least 3.5 million patients worldwide have undergone AIR Recon DL scans.According to the recent AIR Recon DL 3D and PROPELLER reader study:

100% of participants indicated that the deep learning solution provided a better signal-to-noise ratio and better or equivalent image clarity compared to conventional image reconstruction.

99% of people also say that AIR Recon DL provides better or equivalent lesion conspicuity.

The report also shows that the examination time was reduced by 50%, improving the efficiency and productivity of the workflow.

"By extending AIR Recon DL to 3D and PROPELLER, GE Healthcare has narrowed the capability gap in our ability to deliver better image quality and patient experience for all types of examinations, particularly in brain imaging and musculoskeletal imaging where we heavily rely on 3D sequences. PROPELLER is crucial for reducing variability in image quality and eliminating repeat sequences caused by motion.""Dr. Tiron Pechet, radiologist and assistant medical director at Shields Health Care Group, said."

AIR Recon DL and its recent expansion are available for new GE Healthcare MRI scanners and as an upgrade for the majority of 1.5T and 3.0T MRI systems installed by GE Healthcare.，Besides its benefits for the efficiency and productivity of the industry, this upgrade will enable previously installed scanners to be updated and run like the new system - allowing imaging facilities to enjoy state-of-the-art imaging capabilities on their existing systems while saving capital expenditures in today's cost-constrained environment.These include classic models such as HDxt and MR 750. Meanwhile, for traditional 60cm MRI systems sold previously, the Evo upgrade solution has been introduced, allowing an in-place upgrade to the latest large-bore Premier Evo system without replacing the magnet.

In May last year, GE Healthcare, headquartered in Waukesha, Wisconsin, announced that AIR Recon DL, a deep learning image reconstruction technology applicable to all anatomical structures, has received U.S. FDA 510(k) clearance for use on the Signa 7.0T MRI scanner.

GE Healthcare's Deep Learning Image Reconstruction Technology — AIR Recon DL Approved for Use with Its Ultra-High Field SIGNA™ 7T MRI Scanner

Garry Gold, MD, a professor of radiology at Stanford University and a clinical reviewer for GE Healthcare's FDA application, said:These images are truly stunning. "AIR Recon DL for Signa 7.0T fully leverages the advantages of the ultra-high field 7.0T that we love, which is the ability to visualize high-resolution tissue structures, and elevates it to a new level by reducing noise and edge ringing. Therefore, AIR Recon DL with 7.0T helps reveal a more comprehensive picture, providing better clinical insights for improved patient outcomes, and opening new opportunities for research across various fields of care."