The quality of clinical images plays a crucial role in providing accurate diagnoses and optimizing patient care. As technology continues to evolve – deep learning – a subset of artificial intelligence (AI), has emerged as a powerful tool for clinicians. It can positively impact diagnostic accuracy and improve disease detection, which can help improve patient outcomes across clinical specialties.

"Blurry images can compromise clinicians' ability to diagnose with confidence and speed," said Roland Rott, President and CEO, Imaging, GE HealthCare. "At GE HealthCare, we're ushering in a new era of image resolution with deep learning-powered imaging reconstruction. These AI-powered solutions not only enhance image quality but also streamline workflows, enabling radiologists to work more efficiently and cost-effectively. It also has the potential to provide patients’ more timely access to imaging and consequentially may enable earlier diagnosis and treatment. By leveraging AI and deep learning technologies, we remain committed to delivering advancements that help empower clinicians to make more confident diagnoses and improve outcomes for patients worldwide."

New Effortless Recon DL solutions address pain points across care pathways

Across healthcare, there is a need for improved image quality to help early diagnosis and treatment for patients. The need is particularly acute in chronic illnesses, such as neurological conditions, cancer, and cardiovascular disease, where incidence rates are increasing globally due to population growth and aging.ⁱⁱ In fact, 2024 is the first time the U.S. is expected to report more than two million new cancer cases in a year – equaling almost 5,500 cancer diagnoses each day.ⁱⁱⁱ

Imaging is a critical component to addressing these conditions, ensuring early and accurate diagnosis, as well as ongoing management. More than 80% of hospital visits include early diagnostic imaging across over 23,000 conditions,ⁱ underscoring the sheer volume of images. While their work is essential, many radiologists suffer from burnout. According to a survey of more than 8,000 radiology professionals included in a 2024 report from the American Society of Radiologic Technologists (ASRT), more than half (54%) reported feeling emotionally exhausted, and 57% felt underappreciated on the job.^iv

Unveiled at RSNA 2024, Sonic DL for 3D^v is a deep learning innovation designed to reduce MRI scan times by up to 86%. Following the launch of Sonic DL for cardiac imaging, this extension to 3D is expected to offer enhanced resolution for brain, spine, orthopedic, and body imaging – while retaining the same impressive scanning speed of up to 12 times acceleration. For neurology, Sonic DL for 3D is designed to enable high-resolution imaging of complex brain structures, allowing for quicker, clearer insights into neurological conditions.

To assist clinicians in oncology, GE HealthCare’s Clarify DL^vi nuclear medicine solution is designed to enhance bone SPECT image quality, an important factor in increasing diagnostic confidence.^vii In a clinical evaluation, Clarify DL's image resolution was rated as better in 98% of the exams.^viii The AI-powered solution is designed to deliver clear, accurate, and effortless imaging – a stark contrast to today’s noise reduction techniques, that can lower noise at the expense of contrast and resolution. Clarify DL is designed for use with GE HealthCare’s StarGuide SPECT/CT system.

For the growing practice of cardiac CT, GE HealthCare offers TrueFidelity DL with the cardiac kernel, enabling outstanding detail at low doses, with improved visual sharpness for confident reporting and accepted image texture. The value of this AI-powered tool is fully realized when combined with the company’s Revolution Apex platform – which boasts a 0.23 second rotation time for one-beat cardiac imaging – and ECG-less Cardiac solution to acquire cardiac images without the aid of the patients’ ECG signal/trace.

Increasing global access to AI-powered technologies

Sonic DL for 3D, Clarify DL, and TrueFidelity DL each leverage a dedicated Deep Neural Network (DNN), which allows for image reconstruction that produces significantly improved images with better detail and quality. The solutions will now join the company’s extended Effortless Recon DL portfolio, which is bringing critical deep learning advancements to clinicians around the world and includes:

• True Enhance DL (CT) – recently FDA cleared – generates deep learning-based images that can estimate monochromatic 50keV images from single-energy X-ray, to increase contrast and support confident diagnoses. The AI-based solution provides clinicians a simple workflow and incredible image quality. True Enhance DL is now available to new and existing Smart Subscription customers.

• TrueFidelity DL (CT) offers impressive image quality performance^ix, and preferred image sharpness and noise texture,^x without compromising dose performance. As a result, the technology – which also includes kernels for lung and bone imaging – can improve reading confidence across a wide range of clinical applications such as head, whole body and cardiovascular, for patients of all ages. New this year, TrueFidelity DL is now available on Revolution Ascend platform, Revolution Maxima and to new and existing Smart Subscription customers.

"The sharpness of the images is a breakthrough development in image reconstruction algorithms. We see details that we have never seen before. Abdominal, lung and cardiac imaging benefits most from this technology. I am mainly interested in cardiac and cardiovascular imaging. We found much better image quality, depiction of details, and image sharpness for cardiac valves, sclerotic and soft plaque in cardiac and extracardiac vessels, as well as fewer artifacts around stents and stent-grafts. DECT for pulmonary embolism easily convinced everybody in our department." — Prof. Klaus Hergan, University Hospital Salzburg, Austria Department of Radiology of the 1200-bed University Hospital Salzburg^xi

• Precision DL (PET/CT) enhances image quality in PET/CT scans, providing clinicians with a powerful solution to aid in precise diagnoses, treatment planning and monitoring with the image quality performance benefits typically associated with hardware-based Time-of-Flight (ToF) reconstruction, without compromising sensitivity, including improved contrast-to-noise ratio, contrast recovery,^xii and quantitative accuracy.^xiii

• AIR Recon DL (MR and PET/MR) helps improve image sharpness and enables reduced exam times by up to 50% without compromising image quality.^xiv These features help to enhance workflow efficiency and productivity for healthcare professionals, reduce the need for repeat scans, enable faster diagnoses, and enhance patient experience. With expansion from 2D to 3D and PROPELLER imaging sequences, AIR Recon DL expands to nearly all MRI clinical procedures, covering all anatomies.

“The image quality. The scan times, again, probably the biggest revolution we’ve seen in the MRI field in a long time, and I’ve been doing this a long time.” –Tom Schrack, BS, ARMRIT, CS Manager of MR Education and Technical Development Fairfield Radiological Consultants, USA^ix

• Sonic DL (MR), available since 2023, enables cardiac imaging within a single heartbeat, scanning up to 12 times faster compared to conventional methods,^xv enabling rapid cardiac MR functional imaging to match the speed of MRI to the speed of physiology. This minimizes or removes the need for repetitive patient breath holds, simplifying procedures and expanding the pool of patients eligible for cardiac MRI to include arrhythmic patients and those with difficulty holding their breath. At RSNA 2024, GE HealthCare unveiled Sonic DL for 3D^v that will broaden its applicability beyond cardiac.

• Helix Advanced Image Processing (X-ray) revolutionizes diagnostic imaging quality by consistently delivering images with appropriate detail, contrast, and latitude to enable radiologist to visualize key clinical details.

“Particularly in children in the ICU, who are sensitive to ionizing radiation… we were able to reduce the exposure required to achieve a diagnostic quality image… We were able to see through the mediastinum, to still see the bony details of the spine, which indicated that we had sufficient penetration but yet not burn out the details of the lung parenchyma and pulmonary vascular markings, and that really is an indication of a high-quality image.” – Dr. Nghia (Jack) N. Vo, MD Diagnostician-in-chief, Children’s Wisconsin; Chief of Pediatric Radiology, Medical College of Wisconsin^ix

Altogether, GE HealthCare’s Effortless Recon DL represents a collection of innovative deep learning-powered imaging reconstruction solutions available across Imaging modalities. It is designed to elevate imaging quality by optimizing contrast, signal-to-noise ratio, sharpness, and minimizing noise and artifacts to produce exceptionally clear images for improved clinical insights and decision-making.