Intracranial hemorrhage (ICH) can be a “time bomb” if not detected and treated promptly. The actual cost of delays can be measured in monetary terms that impact patients, families, and the health care system. But even more importantly, the cost in terms of loss of function and loss of life can be staggering. CuraCloud has developed software that can detect ICH on non-contrast head CT images in just a few seconds based on a deep learning model. The software can be used to help triage and prioritize reading of images for patients with ICH, thus improving the likelihood of prompt detection and intervention.
CuraCloud’s mission is to collaboratively develop medical AI solutions with healthcare delivery organizations and medical technology leaders to improve diagnostics, care processes, and clinical outcomes. CuraCloud is interested in teaming up with clinical partners to solve their unique clinical challenges using their unique data.
Qi Song, CEO of CuraCloud, shares his insights about how their machine learning and algorithm development activities leverage collaborative projects with industry and technology leaders to improve patient care. He discusses the potential impact the ICH algorithm can have on the morbidity and mortality associated with ICH, and how the ICH triage tool availability can be expanded via the Nuance AI Marketplace for Diagnostic Imaging.
Jonathon Dreyer: Tell us about your business – when and how you started and your development journey.
Qi Song: CuraCloud is a medical AI R&D services company. We collaboratively develop medical AI solutions with clinical partners to improve diagnostics, care processes, and clinical outcomes for healthcare organizations.
Our founding team is a group of senior research scientists and technical leads who have extensive R&D experience in medical image analysis from leading medical imaging companies. In 2016, we secured funding and started CuraCloud. We have recruited more than 15 data scientists and computer vision experts who are interested in applying machine learning to the healthcare industry. Over the past three years, we have developed into a strong professional services organization working with clinical collaborators and device manufacturers all over the world.
JD: What AI algorithms do you have and what do they do?
QS: We have a portfolio of R&D projects including ICH detection, lung nodule detection and characterizations, coronary artery segmentation and stenosis quantification, ultrasound breast cancer classification, chest X-Ray disease classification, digital pathology cancer metastasis detection, and NLP based structured clinical reports. FDA-cleared algorithms will be made available on the Nuance AI Marketplace for Diagnostic Imaging.
JD: What’s the big “Aha” moment when you first show users what your AI algorithm(s) can do for them?
QS: Our clinical partners are impressed by the spectrum and depth of the R&D projects we have demonstrated within the past several years. We have published more than 20 peer-reviewed scientific journal articles and conference papers regarding our AI projects. Our scientists are capable of delivering high-performing machine learning algorithms with state-of-the-art deep learning techniques.
JD: What challenges or needs did you see that drove you to focus on this?
QS: ICH affects over 67,000 people in the USA each year. When patients are being evaluated for ischemic stroke, hemorrhage needs to be ruled out before they are given certain clot-buster drugs. Time is of the essence. With the consolidation of radiology practices, radiologists increasingly have a backlog of scans to read from multiple hospitals. It is important to flag suspected ICH cases as soon as possible so that treatment decisions are not delayed.
JD: What’s the number one benefit you offer?
QS: The number one clinical benefit of using our AI-assisted triage tool is reduced turnaround time (TAT) for ICH patients who need to be treated immediately.
JD: Are there any stories you can share about how your algorithm(s) drove measurable patient care outcomes?
QS: We have carried out a Monte Carlo simulation study to mimic a single radiologist’s one-week worklist for 2,000 times under different clinical settings, in order to quantify the clinical benefits and risks of using AI for optimizing triage prioritization. The study compared the AI-assisted triage with the standard of care prioritization and showed that almost all ICH positive patients have a shorter turn-around time as a result of the prioritization. Notably, “routine” outpatient studies that have positive ICH findings have substantial TAT savings that ranged from several hours to multiple days. In the meantime, we are collaborating with a large radiology group to collect performance data in a clinical setting. We hope to share some clinical study results soon.
JD: What benefits does Nuance and its AI Marketplace for Diagnostic Imaging bring to your users? What problems does the marketplace and integration into Nuance’s workflow solve?
QS: Nuance enjoys a significant market share for its reporting platform and also has an enterprise worklist solution. We wanted to leverage the customer base that Nuance has established and the cloud hosting infrastructure that Nuance offers to shorten the time-to-clinical-use for our solutions.
We also find Nuance’s built-in feedback channel to be a valuable feature that will allow the users of our app to share their results with our team. This will create a communication channel between radiologists and our developers that will enable us to improve our algorithm iteratively over time. We are looking forward to being part of the collaborative community of healthcare developers and users that Nuance has created.
JD: What has your experience been working with the Nuance team?
QS: We are proud to be working with the Nuance team to bring our solutions to radiologists. The Nuance team is incredibly knowledgeable at making innovative healthcare technologies accessible to their customers. We are excited to see the cloud hosting capability that is newly available, and we are eager to be fully integrated and become a part of the AI portfolio of the Nuance AI Marketplace.
JD: What is your vision for how your solution(s) will evolve over the next 5 years?
QS: We plan to expand this specific ICH application from the relatively simple triage functionality to a more comprehensive diagnosis decision support tool. For example, we plan to include more head CT findings, such as hemorrhage subtypes, the hemorrhage location, and measurements, mass effect, and midline shift detection, etc. Beyond that, we have other exciting projects that are ongoing and hopefully will have more significant impacts in the next 5 years. Our main focus is to further expand our professional R&D services to gain more clinical experience by working closely with our clinical partners on their unique needs and challenges by applying state-of-the-art deep learning techniques on their own clinical data.
JD: In one sentence, tell us what you think the future of medicine will look like.
QS: The future of innovation in medicine will be driven by technologies developed in multidisciplinary collaborations that will solve the challenges we face today.
To learn more about CuraCloud, please visit https://CuraCloud.net
To learn more about Nuance AI Marketplace for Diagnostic Imaging, please visit https://www.nuance.com/healthcare/diagnostics-solutions/ai-marketplace.html
Intelligence at Work is a blog series by Jonathon Dreyer, Vice President, Solutions Marketing, Nuance Communications. Intelligence at Work showcases projects and applications that demonstrate how Nuance technologies extend the value, use and performance of integration, and development partner offerings. This blog series focuses on inspiring the healthcare developer community to think beyond their current state and take their innovations to new heights by tapping into the latest in artificial intelligence.