While many imaging professionals recognize and are experiencing the benefits of AI-powered mammography, there can be several questions some imaging centers want answers to before fully embracing it.
In this post, we’ll explore common questions about AI and what real-world radiologists are discovering after adopting AI into their screening workflows.
#1: Why should we move from CAD to AI in reading workflows?
Moving from Computer-Aided Detection (CAD) to AI-powered mammography represents a significant technological advancement in breast imaging. CAD systems, introduced in the late 1990s, were designed to highlight areas of interest on mammograms that radiologists would then review for potential abnormalities. CAD programs weren’t nearly as sensitive or accurate as today’s AI solution, and so radiologists tended to ignore its recommendations.
However, the transition to AI-powered mammography offers several key improvements and benefits:
• Improved accuracy in cancer detection, often before any visible signs
• More efficient workflows and MSQA scores for radiologists
• Reduced reading times and false positives
#2: Can AI be trusted in reading scans?
AI use in breast mammography dates back 20 years, has been extensively studied and validated across global populations and clinical sites, and continuously evolves and improves. Establishing trust in an AI solution, along with changing workflows, is essential toward fully adopting it into any radiology practice. However, if clinicians wait to adjust workflows, or don’t take the time to study and understand how AI is finding hidden cancers, their trust in the solution may be delayed.
#3: Does AI really help radiologists find more cancers?
An AI-solution partner needs to be ready to prove diagnostic accuracy, performance reliability, and demonstrate how their solution improved patient outcomes in clinical studies and real-world facilities. They should be able to remove the “black box” concerns associated with how the algorithm was developed and trained with real world, relevant, patient data. The more AI is used, the more accurate it becomes through its deep learning algorithm, improving the lives of both the clinicians and patients.
Importantly, the deep learning dataset must represent the intended patient population, such as age, race and ethnical population distributions. With the ultimate goal of accurately increasing cancer detection rate by analyzing for masses, distortion, calcifications, and asymmetry, AI localizes, segments and classifies lesions giving them a score and a case score for the overall exam.
#4: Will AI eventually take over and radiologists will lose diagnostic autonomy, or even worse, their jobs?
AI-powered mammography is a partner to radiologists in screening workflows, not a replacement. The discernment and expertise the radiologist brings to the reading is essential to verify any diagnosis or recommendation.
As a radiologist, AI helps you do your job better, faster, and more accurately, enabling you to provide the best care for your patients. Radiologists are finding more of the “hard to find” cancers earlier, and case scores help radiologists triage cases and prioritize their time.
Both in the US and OUS, with single reader and double reader workflows, radiologists use the AI recommendations, lesion scoring, and overall case scoring as a guide to help them prioritize reading on low-to-high-risk cases and identify lesions sooner. AI is an enhancer to clinician judgement and the screening process.
#5: Does AI compromise or help with compliance with regulatory agencies, data privacy, and security standards?
Imaging centers might be concerned about ensuring compliance with regulatory standards when using AI in medical imaging, since adhering to established guidelines and standards is crucial in the healthcare industry. They may also express concerns about the privacy and security of patient data when using AI systems. Ensuring that AI algorithms comply with data protection regulations and standards is crucial to gaining trust.
AI applications in cancer diagnosis must obtain regulatory clearance or approval to ensure the safety and efficacy of these innovative technologies. Typically, regulatory bodies such as the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) play a pivotal role in evaluating and approving AI-based cancer diagnostic tools. Developers must adhere to stringent guidelines, providing comprehensive evidence on the algorithm’s performance, accuracy, and clinical utility. This includes rigorous testing on diverse datasets and robust clinical validation studies. Additionally, developers need to demonstrate adherence to ethical standards, data privacy regulations, and transparent reporting practices.
#6: How quickly can our team learn to use an AI solution?
Implementing AI should always come with a thoughtful onboarding team and process to ensure technical and operational understanding, changes to daily workflow to maximize results, and continuous learning. When considering any AI-powered mammography solution, talk directly with the partner to discuss any concerns about how the solution works, from the algorithm to the user interface. At iCAD , we work closely with the facility and operation teams over a 3-to-6-month period to ensure the solutions fit within a successful workflow, systems are integrated properly to enhance usage by radiologists, and we’re monitoring progress against success measurements. Ultimately, we’re helping ensure you’re prepared to find more cancers, increase efficiencies, and drive team adoption and positive sentiment while learning.
#7: Will the costs outweigh the benefits for our imaging center?
Some of the initial cost considerations when evaluating an AI-powered solution are: licenses and implementation, onboarding training and adoption, and ongoing upgrades and maintenance. While there is always an initial investment to acquire and implement a new solution, it’s crucial to assess it compared to the long-term gains in efficiency, accuracy, and patient outcomes, and these return-on-investment (ROI) benefits.
An AI-powered mammography partner should have a tool to easily demonstrate the ROI of implementing their solution in your radiology facility. The cost-effectiveness provided through earlier, accurate diagnosis, improved cancer detection rates, reduced false positives, and enhanced reputation and referrals will be self-evident.
What are your next steps?
In the next post, we discuss the essential checklist you should follow when evaluating any AI-powered mammography solution and partner. Read More