TechnicalDiscussion

How Does HeartFlow's AI Analyze Your Heart Scan? | Dr Campbell Rogers | EP#426

Dr. Campbell Rogers explains HeartFlow's AI analysis process for heart scans, emphasizing that the company assesses image quality first, rejecting approximately 2-3% of scans that are too fuzzy. He highlights that scanner type, procedure techniques, and medications used at each site significantly impact image quality and the ability to track disease progression over time.

Summary

HeartFlow's AI analysis begins with a critical image quality assessment to ensure scans are sharp enough for detailed AI analysis. According to Dr. Rogers, approximately 97% of images meet the quality threshold for analysis, while the remaining 2-3% are rejected because they are too fuzzy or unclear to provide reliable results.

Dr. Rogers emphasizes that the specific equipment and techniques used at each imaging site make a meaningful difference in scan quality. Each cardiac imaging facility uses its own unique scanner and has its own unique procedures, protocols for scanner operation, and medication administration practices. HeartFlow invests considerable time and resources in working with CT imaging technology to optimize image quality across different sites.

When comparing scans from the same patient over time to assess disease progression, scanner consistency becomes particularly important. Using the same scanner and the same technique for both the initial and follow-up scans improves the reliability of comparisons and the ability to accurately determine whether coronary artery disease has worsened, improved, or remained stable.

Key Insights

  • HeartFlow rejects approximately 2-3% of heart scan images because they are too fuzzy or lack sufficient clarity for detailed AI analysis
  • Each cardiac imaging site operates with its own unique scanner, procedure protocols, and medication administration practices, which impacts image quality outcomes
  • HeartFlow dedicates significant resources to optimizing CT imaging technology to standardize and improve image quality across different sites
  • Using the same scanner and technique for repeat scans in the same patient is important for accurately comparing disease progression over time
  • Scanner consistency and procedural standardization are critical factors in determining whether a patient's coronary artery disease has worsened, improved, or remained unchanged

Topics

Image quality assessment in cardiac AI analysisScanner variability and site-specific proceduresCT imaging optimization techniquesDisease progression trackingScan quality rejection criteria

Transcript

[0:00] How does heart flow perform analysis and how important is the machine that's used for the scan? >> The first thing we do is we do an assessment of image quality. That means images are sharp enough to do our detailed AI analysis. The images are good enough about a little more than 97% and the other two or 3% we don't provide an analysis is the images are just too fuzzy. To your question about the technique at the site, it does make a difference. Each site has its own unique scanner. They have their own unique [0:30] procedure, how the scanner is used, what medications are given. We at HeartFlow spend a lot of time working with…

Full transcript available for MurmurCast members

Sign Up to Access

More from The Proof with Simon Hill

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.