ECG Classification

This project is machine learning algorithm to classify/detect arrhythmia.

I use DNNClassifier in tensorflow.

Arrhythmia Type List

• N = 'N', 'L', 'R', 'e', 'j'

• SVEB = 'A', 'a', 'J', 'S'

• VEB = 'V', 'E'

• F = 'F'

• Q = '/', 'f'

Arrhythmia Annotation

• N = Normal beat

• L = Left bundle branch block beat

• R = Right bundle branch block beat

• e = Atrial escape beat

• j = Nodal (junctional) escape beat

• A = Atrial premature beat

• a = Aberrated atrial premature beat

• J = Nodal (junctional) premature beat

• S = Supraventricular premature or ectopic beat (atrial or nodal)

• V = Premature ventricular contraction

• E = Ventricular escape beat

• F = Fusion of ventricular and normal beat

• / = Paced beat

• f = Fusion of paced and normal beat

The Number of Each Arrhythmia Type

• N = 10001

• L = 8075

• R = 7259

• e = 16

• j = 229

• A = 2546

• a = 150

• J = 83

• S = 2

• V = 7130

• E = 106

• F = 803

• / = 7028

• f = 982

I take types(N, L, R, A, V, /) because quantity of others are not suitable

Environment

---

1. python 3.5.2
2. tensorflow-gpu 1.10.0
3. intel i7-7700K
4. GeForce GTX 1080 x 2

Usage

---

1. Run ECG_Classification.ipynb
2. That's all

Result

---