HMC-QU Dataset

The HMC-QU benchmark dataset is a joint effort by Hamad Medical Corporation (HMC), Tampere University, and Qatar University to advance research in echocardiography and myocardial infarction (MI) detection. Approved by the HMC Hospital ethics board in February 2019, this dataset contains apical 4-chamber (A4C) and apical 2-chamber (A2C) 2D echocardiography recordings collected during 2018 and 2019. These recordings were obtained using ultrasound devices from top vendors, such as Phillips and GE Vivid (GE-Health-USA).

Key Features of the HMC-QU Dataset

High-Quality Echocardiography Data

• Frame Rate: The recordings have a frame rate of 25 frames per second (fps), providing smooth temporal resolution.
• Spatial Resolution: The resolution ranges from 422×636 to 768×1024 pixels, ensuring sharp and detailed visuals for thorough analysis.

Comprehensive Use Cases

1. Myocardial Infarction Detection

   • This dataset is the first publicly available collection dedicated to detecting myocardial infarction (heart attack) in the left ventricle wall.
   • It includes data from over 10,000 echocardiography exams, featuring 800+ cases of acute ST-elevation MI.
   • Ground-truth labels assign each myocardial segment as either MI (indicating regional wall motion abnormality) or non-MI.

2. Left Ventricle Wall Segmentation

   • The dataset facilitates precise segmentation of the left ventricle wall, supporting both clinical diagnostics and machine learning model development.