A Lightweight CNN Net for AMD Detection Using OCT Volumes
Journal
Digest of Technical Papers - IEEE International Conference on Consumer Electronics
Journal Volume
2022-January
ISBN
9781665441544
Date Issued
2022-01-01
Author(s)
Abstract
To popularize age-related macular degeneration diagnosis in rural and remote areas, we proposed a lightweight convolution neural network (CNN) architecture that aims to identify whether the patient has age-related macular degeneration through the optical coherence tomography images. The proposed CNN model achieves 2, 322 parameters and 0.0573 GFLOPs, which is only 4.98% of the Mobile net. Besides, the CNN model achieves 98.18% of accuracy. With the fixed-point simulation using 14 bits on weights and 5 bits on input data, the CNN model achieves accuracy of 97.73 %.
Subjects
Age-related macular degeneration | convolution neural network | deep learning | optical coherence tomography
SDGs
Type
conference paper