Deep learning (DL) technology, a form of artificial intelligence (AI), is advancing medical imaging capabilities by providing more robust, accurate, data-driven information to support more efficient workflows and examinations. increase. Precision DL for PET/CT is the latest technology. GE HealthCare’s Effortless Recon DL portfolio of deep learning-based solutions, which also include AIR Recon DL for MRI, TrueFidelity for CT, and Helix DL for X-ray, significantly improve image quality and clinical information that will help you make decisions.Improving patient outcomes
GE HealthCare (Nasdaq: GEHC) today announced US FDA 510(k) clearance for Precision DL. It is a new innovative deep learning-based image processing software in GE HealthCare’s growing Effortless Recon DL portfolio. Precision DL offers image quality performance advantages typically associated with hardware-based time-of-flight (ToF) reconstruction, such as improved contrast-to-noise ratioi, contrast recoveryi, and quantitative accuracyi. The AI-based technology is available in the company’s fastest-selling PET/CTiii, the Omni Legend. The Omni Legend already boasts more than twice the sensitivity of previous digital scannersiv, enabling faster scan timesv and superior small lesion detection capabilitiesvi.
Availability of Precision DL with Omni Legend ultra-sensitive 3rd generation digital detector technology ushers in a new era in PET/CT performance and results, transitioning from ToF technology to next generation PET/CT performance and can be decoded by the clinician. Detect chance events with very fine resolution for informed diagnosis and treatment planning.
“You can’t treat what you can’t see, so you need accurate image quality to help diagnose, plan and monitor disease,” explains Professor Flavio Forer, President of Nuclear Medicine. Department of Radiology and Nuclear Medicine, St. Gallen-Canton Hospital, Switzerlandvii. “Precision DL improves image quality, enabling the detection of small lesions, such as images obtained with very low dose injections or short sleep times, potentially allowing earlier initiation of treatment and monitoring. Patient outcomes may be improved.” Additionally, the Omni Legend is a streamlined and simple way to help technicians increase efficiency, enhance patient care, and reduce potential radiation exposure for medical staff. provide a solution. ”
Medical imaging is a critical tool for diagnosing disease, identifying treatment strategies, and determining whether treatment has been successful for millions of patients worldwide. Image quality is critical to clinicians and patients, and can make the difference between detecting small lesions early or late, potentially impacting patient outcomes and disease management. For this reason, clinicians are increasingly adopting her AI-based solutions to improve image quality compared to standard care.
Deep learning, a subset of AI and machine learning, makes use of deep neural networks. A deep neural network consists of layers of formulas and millions of connections and parameters that are trained and enhanced based on the desired output. This makes deep learning significantly more effective than previous processes that required more human intervention, and allows clinicians to be more confident because they can easily handle complex models and vast numbers of parameters. Get the time and insight you need to diagnose and care with Patience.
“One of the main benefits of fully moving to an AI and deep learning future is that it will make cutting-edge imaging more accessible to more medical settings than ever before in more medical fields. said Jan Makela, President and CEO of Imaging. , GE Healthcare. “Clinicians are already using the multimodality family of Effortless Recon DL applications, including AIR Recon DL for MR, TrueFidelity for CT, and Helix DL for X-ray, to see the value of applying deep learning technology to improve image quality. We are proud to now add Precision DL for PET/CT, enabling more precise and personalized care across the imaging arm of the medical system.”
More than just a new image processing technique, Precision DL is designed using sophisticated deep neural networks trained on thousands of images produced with multiple reconstruction methods, including ToF reconstruction. , which provides the image quality performance benefits typically associated with hardware-based ToF reconstruction. Improves contrast-to-noise ratio and contrast recovery.
Precision DL processes patient images to improve image quality such as:
-
Average 11% better contrast recoveryi,
-
Average 23% better contrast-to-noise ratioi,
-
an average 42% increase in detection of small, low-contrast lesionsviii, and
-
14% improvement in feature quantification accuracyii.
A study published in the European Journal of Nuclear Medicine and Molecular Imaging demonstrated feature quantification, overall image sharpness, and overall diagnostic value, especially reconstruction without ToF using deep illumination. demonstrated improved lesion detectability and diagnostic confidence on PET/CT images. Learning model ix trained for ToF image enhancement.
GE HealthCare’s deep learning-enabled software is revolutionizing image acquisition and reconstruction in MR, CT, X-ray and now PET/CT, empowering clinicians and helping improve patient outcomes .
For more information on GE HealthCare’s Precision DL, Omni Legend PET/CT, or Effortless Workflow DL, visit gehealthcare.com.
__________________________________
Precision DL using Omni Legend 32cm data improves contrast recovery (CR) by an average of 11% and contrast-to-noise ratio (CNR) by an average of 23% compared to non-ToF reconstruction. CR and CNR were demonstrated using clinical data including implanted lesions of known size, location, and contrast. Using data from Omni Legend 32 cm, CR and CNR were measured using high-precision DL and QCHD.
ii Precision DL with Omni Legend 32cm improves feature quantification accuracy by 14% at comparable noise levels compared to Discovery MI with ToF reconstruction. Quantitative accuracy is demonstrated using clinical data including implanted lesions of known size, location and contrast (ground truth). Compare Omni Legend 32 cm SUVmean with high accuracy DL and Discovery MI 25 cm SUVmean with QCFX.
iii Based on GE HealthCare PET/CT system ordering data from 2010 onwards.
iv Omni Legend 32 cm has up to 2.2 increase in system sensitivity compared to Discovery MI 25 cm. Measurements are in accordance with NEMA NU 2-2018.
v PET scan time was reduced by up to 53% with the Omni Legend 32 cm compared to the Discovery MI 25 cm as demonstrated in the phantom test.
The vi Omni Legend 32 cm improves detection of small lesions by an average of 16% and up to 20% compared to the Scan Time/Injection Volume Matched Discovery MI 25 cm. This has been demonstrated in a phantom test using a model observer with a 4 mm lesion. Average of different reconstruction methods.
vii Not a consultant to GEHC: Statements made by GE customers expressed herein are based on their own opinions and results achieved in the customer’s unique environment. There is no ‘typical’ hospital, there are many variables such as hospital size, case mix, etc., so there is no guarantee that other customers will achieve the same results.
viii Omni Legend 32cm and Discovery MI 25cm were compared to match scan times and injection volumes. Clinical data including an 8 mm diameter liver lesion inserted at a known location and detectability using 2:1 contrast using a CHO model observer. Compare the SNR of Omni Legend 32 cm with QCHD and Precision DL and Discovery™ MI 25 cm with QCFX.
ix Mehranian, A., Wollenweber, SD, Walker, MD et al. Deep learning-based image enhancement of time-of-flight (ToF) non-ToF PET scans. Eur J in Nucl Med Mol Imaging 49, 3740–3749 (2022). https://doi.org/10.1007/s00259-022-05824-7
View source version at businesswire.com: https://www.businesswire.com/news/home/20230530005497/en/
