论文信息 - Predicting Chance of Success on Epiretinal Membrane Surgery using Deep Learning

Predicting Chance of Success on Epiretinal Membrane Surgery using Deep Learning

A preliminary study on predicting chance of success on an epiretinal membrane surgery is studied. Given an optical coherence tomography image, the study shows that the multilayer perceptron neural network can achieve 91.0% accuracy. Due to an unbalance of the images of success and failure classes, under-sampling and over-sampling are applied. For oversampling, the images in the failure class are duplicated to balance the number of images compared to the success class. Utilizing the balance dataset, the prediction performance is improved from 91.0% to 93.0% for over-sampling. With the exploitation of, the salient region for training the model and predicting the outcome. The salient region is manually segmented to express the fovea in the OCT. The experimental results evidence an improvement of 1.0% with achievement of 94.0% accuracy.

[1] M. Akiba,et al. Optical Coherence Tomography-Based Deep-Learning Models for Classifying Normal and Age-Related Macular Degeneration and Exudative and Non-Exudative Age-Related Macular Degeneration Changes , 2019, Ophthalmology and Therapy.

[2] Xiaodong Wu,et al. Automated 3-D Intraretinal Layer Segmentation of Macular Spectral-Domain Optical Coherence Tomography Images , 2009, IEEE Transactions on Medical Imaging.

[3] Adam Dabrowski,et al. Novel full-automatic approach for segmentation of epiretinal membrane from 3D OCT images , 2017, 2017 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA).

[4] James S. Duncan,et al. Medical Image Analysis , 1999, IEEE Pulse.

[5] Jerry L Prince,et al. Retinal layer segmentation of macular OCT images using boundary classification , 2013, Biomedical optics express.

[6] Jie Huang,et al. Efficient Deep Learning-Based Automated Pathology Identification in Retinal Optical Coherence Tomography Images , 2018, Algorithms.

[7] Michael D Abràmoff,et al. A combined machine-learning and graph-based framework for the segmentation of retinal surfaces in SD-OCT volumes. , 2013, Biomedical optics express.