AI and deep learning enable 4x faster scans and productivity gains for clinical radiology

Description

AI and deep learning drive healthcare and clinical radiology innovations. Enhao Gong and Greg Zaharchuk detail FDA-cleared AI solutions that enable faster and safer radiology exams, deliver better care to patients, and provide immediate and quantifiable values to hospitals and imaging centers. The solutions are designed to address issues facing clinical radiology such as the inefficiency and expense of imaging exams (such as MRI and PET), the use of radiation and contrast dosage resulting in risks to patients, and the increasing needs of compacity and the patients’ needs urge radiology practice to improve both quality and productivity.

Subtle Medical provides a deep learning solution, cleared by the FDA and accelerated by industry frameworks such as Intel OpenVINO, to address these problems by enabling 4x–10x faster MRI scans, 4x faster PET scans, and 10x dosage reduction. For faster PET and MRI exams, the deep learning solution is proposed to replace the conventional iterative optimization-based algorithm. Using ResNet-based recurrent structure and generative adversarial network-based adversarial loss, the model successfully reconstructs images from low-quality into high-quality with significantly better signal-to-noise ratio and resolution information. This generates results much more efficiently than conventional iterative methods with much better image quality and accuracy. Based on the deep learning algorithm, Subtle Medical developed SubtlePET and SubtleMR, the first AI software products FDA cleared for medical imaging enhancement. Gadolinium deposition is one of the most urgent issues facing the radiology community. SubtleGad, the deep learning technology, algorithmically restores or boosts the contrast information in MRI. Subtle Medical further verified the generalization and robustness of the DL solution and have shown it to improve workflow while maintaining diagnostic quality.

Deployment with industry solutions such as Intel OpenVINO provides a ~10x speedup of the inference by optimizing the network model. This application, with great accuracy, scalability, and efficiency, can significantly improve the workflow for clinical radiology to provide faster, safer, and smarter imaging exams to patients. Results from clinical partners and deployment sites such as Hoag Hospital and UCSF demonstrate the immediate benefits AI can provide to hospitals and imaging centers. Applications also show the value of the applications to clinical trial and pharmaceutical developments.