Retinal Fluid Segmentation and Classification in OCT Images Using Adversarial Loss Based CNN

This paper proposes a novel method in order to detect the presence and obtain voxel-level segmentation for three fluid lesion types (IRF/SRF/PED) in OCT images provided by the ReTOUCH challenge. The method is based on a deep neural network consisting of encoding and de-coding blocks connected with skip-connections which was trained using a combined cost function comprising of cross-entropy, dice and adversarial loss terms. The segmentation results on a held-out validation set shows that the network architecture and the loss functions used has resulted in improved retinal fluid segmentation.

[1]  A D Négrel,et al.  2002 Global update of available data on visual impairment: a compilation of population-based prevalence studies , 2004, Ophthalmic epidemiology.

[2]  Peter K Kaiser,et al.  Optical coherence tomography imaging of macular oedema , 2014, British Journal of Ophthalmology.

[3]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[4]  Bianca S. Gerendas,et al.  Correlation of 3-Dimensionally Quantified Intraretinal and Subretinal Fluid With Visual Acuity in Neovascular Age-Related Macular Degeneration. , 2016, JAMA ophthalmology.

[5]  Michael F. Marmor,et al.  Mechanisms of fluid accumulation in retinal edema , 2004, Documenta Ophthalmologica.

[6]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[7]  Bianca S. Gerendas,et al.  Joint retinal layer and fluid segmentation in OCT scans of eyes with severe macular edema using unsupervised representation and auto-context. , 2017, Biomedical optics express.

[8]  Sina Farsiu,et al.  Kernel regression based segmentation of optical coherence tomography images with diabetic macular edema. , 2015, Biomedical optics express.

[9]  Yue Wu,et al.  Deep-Learning Based, Automated Segmentation of Macular Edema in Optical Coherence Tomography , 2017, bioRxiv.

[10]  Linda Drake.  Prevention of Blindness from Diabetes Mellitus - Report of a WHO ConsultationPrevention of Blindness from Diabetes Mellitus - Report of a WHO Consultation , 2007 .

[11]  Milan Sonka,et al.  Three-Dimensional Analysis of Retinal Layer Texture: Identification of Fluid-Filled Regions in SD-OCT of the Macula , 2010, IEEE Transactions on Medical Imaging.

[12]  Yue Wu,et al.  Deep-Learning Based, Automated Segmentation of Macular Edema in Optical Coherence Tomography , 2017, bioRxiv.

[13]  Liang Liu,et al.  Automated volumetric segmentation of retinal fluid on optical coherence tomography. , 2016, Biomedical optics express.

[14]  Xinjian Chen,et al.  Three-Dimensional Segmentation of Fluid-Associated Abnormalities in Retinal OCT: Probability Constrained Graph-Search-Graph-Cut , 2012, IEEE Transactions on Medical Imaging.

[15]  Nassir Navab,et al.  ReLayNet: Retinal Layer and Fluid Segmentation of Macular Optical Coherence Tomography using Fully Convolutional Network , 2017, Biomedical optics express.

[16]  Camille Couprie,et al.  Semantic Segmentation using Adversarial Networks , 2016, NIPS 2016.