A fused deep learning architecture for viewpoint classification of echocardiography

•An improved CNN architecture achieving classification result of 92.1% precision.•Spatial CNN network delivers 89.5% precision for the eight classes.•2D KAZE and 2D KAZE+Optical Flow give precision of 89.4% and 84.3% respectively.•3D KAZA and Optical Flow provide precision rates of 87.9% and 79.4% respectively.•2D SIFT and 3D SIFT realize accuracy of 83.8% and 73.8% respectively.