Medical imaging has revolutionized disease diagnosis, yet the potential is hampered by limited access to diverse and privacy-conscious datasets. Open-source medical datasets, while valuable, suffer from data quality and clinical information disparities. Generative models, such as diffusion models, aim to mitigate these challenges. At Stanford, researchers explored the utility of a fine-tuned Stable Diffusion model (RoentGen) for medical imaging data augmentation. Our work examines specific considerations to expand the Stanford research question, Could Stable Diffusion Solve a Gap in Medical Imaging Data? from the lens of bias and validity of the generated outcomes. We leveraged RoentGen to produce synthetic Chest-XRay (CXR) images and conducted assessments on bias, validity, and hallucinations. Diagnostic accuracy was evaluated by a disease classifier, while a COVID classifier uncovered latent hallucinations. The bias analysis unveiled disparities in classification performance among various subgroups, with a pronounced impact on the Female Hispanic subgroup. Furthermore, incorporating race and gender into input prompts exacerbated fairness issues in the generated images. The quality of synthetic images exhibited variability, particularly in certain disease classes, where there was more significant uncertainty compared to the original images. Additionally, we observed latent hallucinations, with approximately 42% of the images incorrectly indicating COVID, hinting at the presence of hallucinatory elements. These identifications provide new research directions towards interpretability of synthetic CXR images, for further understanding of associated risks and patient safety in medical applications.

Large-scale medical imaging datasets have accelerated development of artificial intelligence tools for clinical decision support. However, the large size of these datasets is a bottleneck for users with limited storage and bandwidth. Many users may not even require such large datasets as AI models are often trained on lower resolution images. If users could directly download at their desired resolution, storage and bandwidth requirements would significantly decrease. However, it is impossible to anticipate every users' requirements and impractical to store the data at multiple resolutions. What if we could store images at a single resolution but send them at different ones? We propose MIST, an open-source framework to operationalize progressive resolution for streaming medical images at multiple resolutions from a single high-resolution copy. We demonstrate that MIST can dramatically reduce imaging infrastructure inefficiencies for hosting and streaming medical images by >90%, while maintaining diagnostic quality for deep learning applications.

As a mechanical engineering graduate student, Alison Marsden studied how to make submarines more stealthy. Moving through the ocean, submarines make sounds that can reveal their location. While earning her PhD at Stanford University in the early 2000s, Marsden conducted U.S. Navy-funded research that used sophisticated computer modeling to optimize the shape of the submarines' hydrofoils, which work like airplane wings, generating lift and stabilizing the submarine underwater. Her aim: to minimize telltale churning sounds and enable the vessels to cruise subsurface, undetected. Marsden has always loved the science of fluid mechanics and she enjoyed the technical aspects of her submarine research, but she knew national defense work would not sustain her interest long-term.

This blog post is co-authored by Vittorio Accomazzi, Chief Technical Officer (CTO) at IMS. This blog is part of a series in collaboration with our partners and customers leveraging the newly announced Azure Health Data Services. Azure Health Data Services, a platform as a service (PaaS) offering designed to support Protected Health Information (PHI) in the cloud, is a new way of working with unified data--providing care teams with a platform to support both transactional and analytical workloads from the same data store and enabling cloud computing to transform how we develop and deliver AI across the healthcare ecosystem. The first implementation of digital imaging techniques in clinical use started in the 1970s. Since then, the medical imaging industry has grown exponentially--over the last two and a half decades, there has been a significant development in image acquisition solutions, which has boosted image quality and adoption in different clinical applications.

A recent report by ProPublica showed that medical imaging data of over 5 million patients in the United States are publicly available on the internet. As a result of 187 misconfigured servers, medical imaging data, often containing identifiable patient information that should be protected, is "sitting unprotected on the internet and available to anyone with basic computer expertise." Researchers discovered over 13.7 million medical tests, including 400,000 with downloadable images. These imaging records were stored on servers, including systems used for archiving medical images, without a robust solution in place to monitor for unauthorized changes or to ensure the servers were securely configured and in compliance with regulatory standards. These medical images include MRI, X-Rays and accompanying identifiable patient data that could be used for blackmail.

Imagine the amount of medical imaging data being collected every day in the United States through various equipment,... You forgot to provide an Email Address. This email address doesn't appear to be valid. This email address is already registered. You have exceeded the maximum character limit.

Israeli machine-learning radiology firm Zebra Medical Vision said today it raised 12 million to support the development of imaging algorithms being designed for automatic reading and diagnosis of medical imaging data. The round was led by InterMountain Healthcare, and joined by existing investors, Zebra Medical Vision said. As part of its investment, InterMountain Healthcare plans to work with Zebra to accelerate its development. "We are privileged that 1 of the top healthcare systems in the U.S. has placed such confidence in our team and our platform. In an environment where computing power and machine learning frameworks are becoming a commodity, the ability to quickly and efficiently curate large quantities of data from a world class integrated healthcare provider can make the difference between simplistic tools and insights that can truly add clinical value and positively impact patient care," CEO Elad Benjamin said in a prepared statement.