Related Experiment Videos

Classification of left ventricular thrombi by their history of systemic embolization using pattern recognition of

American Heart Journal
|October 1, 1985
PubMed

Insights

Predicting peripheral embolization from left ventricular (LV) thrombi is crucial. Texture features from two-dimensional echocardiography (2DE) combined with clinical data can accurately identify patients at high risk for emboli.

Area of Science:

  • Cardiology
  • Medical Imaging
  • Artificial Intelligence in Medicine

Background:

  • Diagnosing left ventricular (LV) thrombi is possible with two-dimensional echocardiography (2DE).
  • Factors predicting peripheral embolization from LV thrombi require further delineation.
  • Understanding these factors is critical for preventing embolic events.

Purpose of the Study:

  • To identify clinical and echocardiographic variables associated with peripheral embolization in patients with LV thrombi.
  • To develop a predictive model for embolization risk using texture features from 2DE and clinical data.

Main Methods:

  • Retrospective analysis of 38 patients with LV thrombi diagnosed by 2DE.
  • Inclusion of clinical variables and 2DE texture features (e.g., run length, gray level statistics).
  • Utilized an expert system (Expert Ease) for classification into emboli vs. no emboli groups.

Main Results:

  • Eight patients experienced clinically apparent emboli (leg or brain) within a week of 2DE.
  • Key predictive variables included run length, long runs emphasis, gray level difference statistics, gray level second-order statistics, and warfarin status.
  • The developed model demonstrated a predicted accuracy of at least 96%.

Conclusions:

  • Specific 2DE texture features and clinical data can accurately predict peripheral embolization in LV thrombi patients.
  • This predictive model offers a valuable tool for risk stratification and management.
  • Early identification of high-risk patients can guide preventative strategies.

Related Concept Videos