Intelligent Eye And Vision Health-care System Using Deep Neural Networks

Choroid layer is a vascular layer of tissues between retina and sclera.Optical Coherence Tomography(OCT)is a main imaging technology to image this retinal feature.Various studies have proven that the thickness of the choroid layer is a main attribute factor used for the diagnosis of several ophthalmic diseases.Despite the contemporary advances,automatic segmentation of the choroid layer(i.e.,thickness)is still a challenging task because of the inherent nature of low contrast of OCT images.Despite contemporary advances,automatic segmentation of the choroid layer remains a challenging task due to the low contrast,inhomogeneous intensity,inconsistent texture and ambiguous boundaries between the choroid and sclera in Optical Coherence Tomography(OCT)images.The majority of currently implemented methods manually or semi-automatically segment out the region of interest.While many fully automatic methods exist in the context of choroid layer segmentation,more effective and accurate automatic methods are required in order to employ these methods in the clinical sector.So based on the requirement this thesis proposed and implemented three different methods in the context of choroid layer segmentation and its thickness map.The first method proposed in this context made use of graph cut.The proposed method divided the image into several patches and ran the normalized cut on every image patch separately.The aim was to avoid insignificant vertical cuts and focus on horizontal cutting.After processing every patch,we acquired a global cut on the original image by combining all the patches.Later we measured the choroidal thickness which is highly helpful in the diagnosis of several retinal diseases.The results were computed on a total of 525 images of 21 real patients.Experimental results showed that the mean relative error rate of the proposed method was around 0.4 as the compared the manual segmentation performed by the experts.As the main objective of thesis was to make use of deep learning,we further explored the topic based on the deep learning methodologies.The method proposed and implemented an automatic method for choroid layer segmentation in OCT images using deep learning and a series of morphological operations.The aim of this research was to segment out Bruchs Membrane(BM)and choroid layer to calculate the thickness map.BM was segmented using a series of morphological operations,whereas the choroid layer was segmented using a deep learning approach as more image statistics were required to segment accurately.Several evaluation metrics were used to test and compare the proposed method against other existing methodologies.Experimental results showed that the proposed method greatly reduced the error rate when compared with the other state-of-the-art methods.The proposed method further increase the segmentation accuracy.After moving into the field of deep learning,we proposed a novel choroid layer segmentation method based on the combination of deep learning and hand-crafted features to further increase the segmentation accuracy.For image segmentation,handcrafted features contribute domain related knowledge whereas CNN methods are related with the massive size of general feature sets.There is a plea to merge these two different classes of feature generation methods.The challenge is to form a combined set of features that can possibly outperform either class of feature extraction methods independently.We proposed a cascaded method for choroid layer segmentation that logically combines a CNN feature set with handcrafted features.Our method used as the handcrafted features,Gabor features,Haar features,and gray-level co-occurrence features due to the robustness to segment low contrast images.A Support Vector Machine(SVM)was independently trained using either the CNN feature set or handcrafted feature set,which were than linearly combined for the final segmentation of the choroid layer.The proposed method was evaluated on a data-set of 525 images 21 clinical patient studies from Shanghai Sixth Peoples Hospital.In addition,we proposed two metrics to quantitatively measure the layer thickness:(i)the pixel-wise error in the segmentation;and(ii)an average error in the thickness map being generated.The experimental results showed that the proposed method achieved an accuracy of 98 percent with a mean error rate of 2.84 and outperformed existing state-of-the art segmentation methods.