Fig. 1From: Automated detection of myopic maculopathy from color fundus photographs using deep convolutional neural networksThe framework of our proposed DCNN approach. a The processed images have more uniform color histogram distribution and better clarity than the original images in most cases. b Brief structure of the DCNN-DS model using both original and processed images as inputs. c The classification output into no MM, TF, or PM. MM, myopic maculopathy; TF, tessellated fundus; PM, pathologic myopia; CHDO, color histogram distribution optimization; Conv, convolution; MBConv, mobile inverted bottlrneck convolution; Concat, concatenation; GAP, global average pooling; FC, full connectionBack to article page