Challenge background
Heart failure is a worldwide pandemic that affects at least 26 million people and is becoming more common. Heart failure continues to be a significant public health issue with rising expenses. A crucial indicator for the diagnosis and treatment of heart failure is the ejection fraction (EF). Ejection fraction measures how much blood leaves your heart with each contraction. It is only one of many tests that your doctor might use to find out how well your heart functions. The ejection fraction is often solely tested in the left ventricle. The heart's left ventricle serves as its primary pumping chamber. It forces oxygen-rich blood up into the aorta, the main artery in your body, to supply the rest of your body.
Currently, the gold standard for determining left ventricular ejection fraction is cardiovascular magnetic resonance imaging (CMR). However, each cardiac MRI scan can cost anywhere between $100 and $5,000, or 5.5 times as much as an echocardiogram. Therefore, switching from CMR to echocardiograms to measure left ventricular ejection fraction would have significant health and financial benefits.
The problem
Echocardiography is critical in cardiology. However, the full promise of echocardiography for precision medicine has been constrained by the requirement for human interpretation. In addition, the sonographer's experience is crucial for the human evaluation of heart function, and despite their years of training, there remains inter-observer variability. Echocardiograms have a complex multi-view format, which contributes to the fact that deep learning, a new method for image analysis, has not yet been widely used to analyze them. An artificial intelligence (AI) solution can help identify cardiac structures with accuracy and automate LVEF measurement and myocardial motion with confidence.
Goal of the project
The project's primary purpose is to accurately predict LVEF measurement. With a duration of 4-weeks, this project aims to:
- Data Collection and Exploratory Data Analysis
- Preprocessing
- Feature Extraction
- Model Development and Training
- Evaluate Model
Project timeline
- 1
Week 1
Research previous work and Data Collection
- 2
Week 2
Data Collection
- 3
Week 3
Exploratory Data Analysis
- 4
Week 4
Preprocessing and Augmentation
- 5
Week 5
Model Development
- 6
Week 6
Model Training
- 7
Week 7
Analysis
- 8
Week 8
Streamlit or app
What you'll learn
- Medical image processing
- Computer Vision
- Biomedical Image Analysis