Deep RetinaNet for Dynamic Left Ventricle Detection in Multiview Echocardiography Classification

Figure 3

We input test images into our trained RetinaNet for independent testing and automatically identified A2C, A3C, and A4C images and detected the localization of the left ventricle. The type of view the network predicts is shown in the upper left corner with white text. The red rectangle in the middle represents the left ventricle position detected by our model. Left ventricle detection results in (a) A2C, (b) A3C, and (c) A4C.

(a)

(b)

(c)