AI in Medicine: Medical Imaging Reimagined
ORAL · Invited
Abstract
Medical imaging, a cornerstone of healthcare, uses modalities like X-rays, Computer Tomography (CT), Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET) and ultrasound (US) to diagnose, treat, and monitor diseases. Despite its transformative impact, traditional imaging faces challenges such as high costs, physics/hardware limitations affecting image quality, and limited accessibility in resource-constrained settings. The integration of artificial intelligence (AI) is reimagining medical imaging by addressing these limitations and enabling advancements in accuracy, efficiency, and scalability.
Harnessing the power of AI and Deep Learning (DL), the image transformation model can be trained to predict the target images from the reference inputs. A generalized neural network model that decodes and encodes the image pixels will be presented.
AI’s role in medical imaging spans diverse applications. Within individual imaging modalities, AI enhances image acquisitions & reconstructions, artifact reduction, registrations, and segmentation processes, supporting diagnostic radiologists in detecting subtle anomalies with greater precision. These tools streamline workflows, reduce diagnostic errors, and allow clinicians to focus on complex cases.
AI also transforms imaging across modalities, such as generating synthetic CT images from MRI scans for radiation therapy planning in Radiation Oncology. For PET/MR scanners, MR-based synthetic CT enables Attenuation Correction, crucial for any PET reconstruction methods. Emerging CT-based synthetic MRI offers the soft tissue contrasts for the cancer treatment planning in MR resource-limited Radiation Oncology clinics. Integration of multimodal imaging data provides comprehensive insights, advancing personalized medicine and telehealth applications.
This symposium will explore these innovations while addressing ethical challenges with AI in medical imaging. We are aiming to inspire future applications of AI in medical imaging, democratize access to cutting-edge diagnostics and cancer treatment, and reimagine the future of medical imaging for improved patient outcomes.
Harnessing the power of AI and Deep Learning (DL), the image transformation model can be trained to predict the target images from the reference inputs. A generalized neural network model that decodes and encodes the image pixels will be presented.
AI’s role in medical imaging spans diverse applications. Within individual imaging modalities, AI enhances image acquisitions & reconstructions, artifact reduction, registrations, and segmentation processes, supporting diagnostic radiologists in detecting subtle anomalies with greater precision. These tools streamline workflows, reduce diagnostic errors, and allow clinicians to focus on complex cases.
AI also transforms imaging across modalities, such as generating synthetic CT images from MRI scans for radiation therapy planning in Radiation Oncology. For PET/MR scanners, MR-based synthetic CT enables Attenuation Correction, crucial for any PET reconstruction methods. Emerging CT-based synthetic MRI offers the soft tissue contrasts for the cancer treatment planning in MR resource-limited Radiation Oncology clinics. Integration of multimodal imaging data provides comprehensive insights, advancing personalized medicine and telehealth applications.
This symposium will explore these innovations while addressing ethical challenges with AI in medical imaging. We are aiming to inspire future applications of AI in medical imaging, democratize access to cutting-edge diagnostics and cancer treatment, and reimagine the future of medical imaging for improved patient outcomes.
–
Publication: In preparation
Presenters
-
Yutaka Natsuaki
University of Southern California
Authors
-
Yutaka Natsuaki
University of Southern California