{
  "abstract": "Introduction Trastuzumab is associated with cardiotoxicity, necessitating interval cardiac monitoring. Our centre utilises cardiac magnetic resonance (CMR) measured ejection fraction (EF) for serial heart function assessment. This pilot study evaluated the accuracy and reliability of an AI-powered method using Global Longitudinal Strain (GLS) and long-axis EF (LA EF) compared to conventional volume-derived EF (SA EF) from CMR in patients undergoing Trastuzumab treatment for breast cancer.Methods We retrospectively analysed CMR imaging of five patients who had serial scans as per protocol between 2021–2024. AI-assisted software (Q Mass) generated GLS and LA EF from three long-axis views (4-chamber, 3-chamber, and 2-chamber), which were compared to SA EF derived manually from CMR, the gold standard for clinical decisions.Results Five patients underwent 34 scans during the monitoring period. Bland-Altman analysis showed a linear regression with no significant proportional bias between SA EF and LA EF (p = 0.377). GLS and SA EF demonstrated a strong inverse correlation (Spearman coefficient = 0.77, p < 0.0001) with a worsening GLS (i.e. becoming less negative) corresponding to a lower EF. ROC analysis indicated that GLS had a moderate ability to detect SA EF < 50% (AUC = 0.667).Conclusion EF derived by LA EF and SA EF methods showed consistent differences, supporting the reliability of AI-generated LA EF for serial LV function assessment. AI-based LA EF provides a rapid and reproducible monitoring method, reducing interobserver variability. While GLS and EF are strongly correlated, GLS shows moderate predictive ability for EF decline below 50%. Further studies are warranted to enhance its clinical utility.Conflict of Interest None",
  "authors": [
    {
      "affiliations": [
        "Trent Cardiac Centre, Nottingham University Hospitals NHS Trust"
      ],
      "name": "C Gregory"
    },
    {
      "affiliations": [
        "Trent Cardiac Centre, Nottingham University Hospitals NHS Trust"
      ],
      "name": "S Qureshi"
    },
    {
      "affiliations": [
        "Trent Cardiac Centre, Nottingham University Hospitals NHS Trust"
      ],
      "name": "J Yeo"
    },
    {
      "affiliations": [
        "Trent Cardiac Centre, Nottingham University Hospitals NHS Trust"
      ],
      "name": "S Rekhraj"
    },
    {
      "affiliations": [
        "Trent Cardiac Centre, Nottingham University Hospitals NHS Trust"
      ],
      "name": "M Sosin"
    },
    {
      "affiliations": [
        "Trent Cardiac Centre, Nottingham University Hospitals NHS Trust"
      ],
      "name": "T Mathew"
    }
  ],
  "title": "6-010 AI-driven global longitudinal strain and long-axis ejection fraction by cardiac MRI provide a better assessment to serially monitor heart function compared to manually derived short axis EF in breast cancer patients on trastuzumab",
  "uid": "dadf49da-5bb3-5785-b287-2f1038c8cc50"
}