Corneal Endothelium Image Segmentation Using Feedforward Neural Network

Abstract

Abstract-In this paper the problem of corneal endothelium image segmentation is considered. Particularly, a fully automatic approach for delineating contours of corneal endothelial cells is proposed. The approach produces one pixel width outline of cells. It bases on a simple feedforward neural network trained to recognize pixels which belong to the cell borders. The edge probability (edginess) map output by the network is next analysed row by row and column by column in order to find local peaks of the network response. These peaks are considered as cell border candidates and in the last step of the method via binary morphological processing are linked to create continuous outlines of cells. The results of applying the proposed approach to publicity available data set of corneal endothelium images as well as the assessment of the method against ground truth segmentation are presented and discussed. Obtained results show, that the proposed approach performs very well. The resulting mean absolute error of cell number determination is around 5% while the average DICE measure reaches 0.83 which is a good result, especially when one pixel width objects are compared.

Fully convolutional architecture vs sliding-window CNN for corneal endothelium cell segmentation

“…Regarding the use of neural networks or CNNs to segment CE images, four algorithms were published in the last year. Fabijańska [ 29 ] proposed a feed-forward neural network with one hidden layer to segment 30 ex vivo endothelial images from phase-contrast microscopy (dataset published in [ 30 ]), achieving an error in cell number detection of 5% and a DICE [ 31 ] value of 0.85. Nurzynska [ 32 ] further improved the results on the same dataset by employing a CNN in a sliding-window setup, using a similar network as Cireşan et al [ 3 ], and obtaining a precision of 93% and a DICE of 0.94.…”

A fully automated cell segmentation and morphometric parameter system for quantifying corneal endothelial cell morphology

“…For instance, Katafuchi and Yoshimura [17] proposed a new segmentation algorithm based on Convolutional Neural Network (CNN) to detect the cell contours regardless to the scale of cells. Fabijanska [18] developed an efficient algorithm to address the problem of corneal endothelium image segmentation using Feed-Forward Neural Network (F-FNN), trained to recognize pixels whether they belong to the cell borders or not. However, the main issue of using the supervised segmentation approaches is the time required for training the neural network in order to be able to accurately detect the cell contours.…”

“…The proposed CEAS system is a fully-automated system which requires no user intervention to accurately detect the cell contours. Unlike, other supervised segmentation approaches [17] [18], which require a long time for training the neural network to detect the cell contours, no training procedure is required using the proposed CEAS system. It also enables the quantification of the additional morphometric features (e.g.…”

Four Categories of Human Teeth Based on Biogeography-based Optimization Algorithm and Multilayer Perceptrons

“…B. Multilayer Perceptron MLP is a feed-forward neural network [19][20][21], it usually consists of three parts: input layer, one or more hidden layers and output layer. Its basic model shown in FIGURE III is as follows (a hidden layer).…”