By Erin Register
The Radiological Society of North America (RSNA) held its annual assembly and meeting this past December 1-6 in Chicago, Illinois at McCormick Place. The theme of the assembly was “See Possibilities Together,” and while the week was full of innovative science and quality education, the Artificial Intelligence (AI) Showcase was one of the biggest highlights.
Attendees were able to see AI in action with presentations on the latest topics in AI, machine learning and deep learning. Sponsored by Zebra Medical, the showcase featured over 130 companies, offering attendees the opportunity to experience AI software and product demonstrations. They could also connect with industry leaders. It included several different workshops and sessions, such as an “AI Challenge,” “AI Hands-On,” “RSNA AI Deep Learning Lab” and “Special Interest Sessions.”
This year, the “AI Challenge” was intracranial hemorrhage detection. Attendees were able to view the most successful entries from the 2019 challenge. RSNA worked with four contributing institutions to assemble a dataset of over 25,000 brain CT exams and with volunteers from the American Society of Neuroradiology to label the exams for the presence of five types of intracranial hemorrhages.
In addition to the “AI Challenge,” the showcase also featured an “AI Hands-on Workshop,” where attendees engaged with AI vendors and were able to interact with their systems, using their own laptops in a classroom environment. Exhibitors offered a 90-minute session of user training and product instruction on the latest innovations.
The special interest sessions focused on advanced and innovative radiology topics. One of the sessions was titled “The Role of AI in Radiology in the Developing World,” where presenters discussed various unique programs that are assisting radiologists in low-resource countries. These discussions included the use of artificial intelligence (AI) in providing specific radiologic diagnoses and treatments, including tuberculosis, women’s imaging and diabetic retinopathy.
The “AI Deep Learning Lab” featured four unique sessions developed by RANA members focusing on using open-source tools to complete DL tasks. Sessions included an introductory course focusing on the basic concepts of convolution neural networks (CNNs), a session focused on the use of DL methods for image segmentation and a session describing a recent advance of DL known as Generative Adversarial Networks. Attendees were invited to bring their own devices to begin completing actual tasks in DL.
Several companies at the AI Showcase shared new ideas and products to optimize the workflow of radiologists. Care Mentor AI, an international company specializing in AI applications in medical image analysis, was one of these companies. At their booth, Care Mentor showed attendees the capabilities of their “out-of-the-box” diagnostic systems intended for the interpretation of X-ray images of the chest, foot, knee joint and breast. They also announced their products that are under development, including neural networks for interpretation head and chest CT results. Additionally, Care Mentor offered various clinical and business scenarios demonstrating how their solutions can be applied in medical institutions.
Care Mentor AI Co-founder Pavel Roytberg stated, “Advanced technology enables the optimization of the work of radiologists across the world. Not only does it increase the level of interpretation precision carried out by doctors, but it also saves time and makes interpretations more cost-effective, which certainly impacts the overall performance of the medical institution and, thereby, makes high-quality medical assistance more available.”
Another company to showcase at the AI exhibit was Lunit, a medical AI software company. Lunit presented its latest AI software for radiology — Lunit INSIGHT CXR and Lunit INSIGHT MMG. The Lunit INSIGHT CXR, which recently gained its CE mark, is the most up-to-date AI solution for chest X-ray, and the Lunit INSIGHT MMG is intended for mammography. According to Lunit, the chest X-ray software has been highlighted recently in major publications such as Radiology, Scientific Reports and JAMA Open Network.
Lunit CEO Brandon Suh said, “Every year, we see the AI pavilion enlarged in scale and importance. We are happy to see that the acceptance of advanced technology within the medical society is being amplified.”
CureMetrix, a healthcare technology company that develops AI-driven software for radiology, was also among the 130 vendors at the AI Showcase. Recently named “Best New Radiology Vender” in October by Aunt Minnie Radiology Experts, CureMetrix received FDA-clearance for cmTriage, the first FDA-cleared workflow optimization software in the U.S. that leverages the power of artificial intelligence for breast cancer screening. It enables radiologists to triage, sort and prioritize their mammography worklist based on cases that may need immediate attention.
AI was definitely the prominent subject of discussion at this year’s RSNA Annual Meeting. Change Healthcare noted that “if you properly integrate AI, it will improve the workflow.” Radiology can often be about comparison, and AI adds context to these comparisons to improve patient safety, and overall satisfaction.
For more information on the AI showcase, visit rsna.org/annual-meeting.