The transformative power of AI in the realm of medical imaging has been the subject of extensive discussion for over a decade. However, the practical implementation of AI algorithms in this field, although promising, is laden with challenges. The FDA has, to date, approved over 500 AI algorithms for medical imaging, but transforming these algorithms into affordable, usable products remains a hurdle for companies. Furthermore, the integration of these tools into the imaging workflow often introduces friction, thus inhibiting widespread adoption.
A paper published in Frontiers in Cardiovascular Medicine in 2019 emphasized the potential for medical imaging AI to enhance the quality, equality, and cost-effectiveness of healthcare systems. The authors also forecasted other benefits such as improved patient-physician relationships, better healthcare delivery, and increased physician job satisfaction. Yet, they also acknowledged the challenges AI implementation faces, predicting its initial application in well-circumscribed tasks, with the ultimate goal of integrating these tasks into a seamless and efficient pipeline.
The ideal scenario is for AI integration into the existing workflow to be seamless, leading to improved outcomes as the only noticeable difference from a consumer standpoint. To achieve this goal, it’s crucial to address several key obstacles: earning the physicians’ trust in AI tools, enhancing accessibility, ensuring transparency, and granting physicians the autonomy to decide whether to utilize or dismiss AI-derived information.
As of early 2023, cardiology-related AI algorithms rank second among imaging specialties in the number of FDA-approved AI algorithms. Many companies are developing solutions in the cardiac imaging AI space, but navigating these offerings and attaining adoption is a complex, expensive, and a time-sensitive process. When a hospital wishes to implement an AI tool, it typically entails a lengthy period of research, evaluation, purchasing and IT approvals, and resource allocation for deployment. Each selection of a new AI vendor triggers this process anew.
Upon selection, an AI tool often operates in isolation or remains disconnected from an institution’s daily imaging workflow. Most AI algorithms for cardiac imaging today are either tied to a specific vendor or exist as separate software, which poses a challenge for busy cardiology departments that serve hundreds of patients a day. Therefore, to drive adoption, integrating the tool into the standard imaging workflow is critical.
Transparency is another crucial factor in building physician trust and promoting the adoption of AI tools. As Seetharam, Shrestha, and Sengupta stated, machine learning (a subset of AI) shows promising results in cardiac imaging by improving decision-making based on identified data patterns. Deep learning, inspired by the human brain’s processing capability, takes this a step further. As these technologies advance, providing transparency into what tools are being used to review and evaluate cardiac imaging becomes imperative for both physicians and patients.
For widespread use, AI tools must be readily accessible and offer physicians the control to decide where and how to apply the resultant information.
The future of AI in cardiac imaging, like many new technologies, is teeming with innovative tools. The challenge for the medical imaging industry is to ensure these tools enhance practitioners’ workflow rather than impede it, while also facilitating broad availability to physicians in institutions of various sizes, private practices, and imaging centers.
The onus may fall on the existing medical imaging industry players to resolve these challenges. They could draw inspiration from other industries that have seamlessly integrated emerging technologies into everyday life, such as Amazon. Starting as a platform for books, Amazon has evolved into a streamlined, user-friendly platform where customers can compare options, enhancing their trust and reliance on the technology. Consequently, a plausible direction for AI in cardiac imaging could be the creation of a platform that simplifies access to AI tools as much as placing a one-click Amazon order or purchasing through Apple Pay. The potential impact of such an innovation on facilities of all sizes and patient care is significant.
This article was written by Laurie Smith, and published in the August 2023 issue of Echo Magazine.
Laurie Smith is a principal and COO at Core Sound Imaging, Inc.—makers of the Studycast System, a comprehensive imaging workflow platform. The Studycast System is a platform for medical imaging workflow that has been disrupting and streamlining medical image storage and reporting for 15 years. Studycast is connecting physicians to their images and interpretation tools from any Internet-connected device.
 Artificial Intelligence Will Transform Cardiac Imaging—Opportunities and Challenges
 Karthik Seetharam, Sirish Shrestha, Partho P Sengupta, Artificial Intelligence in Cardiac Imaging, US Cardiology Review 2019;13(2):110–6.