The retinal vascular tree (RVT) is crucial for the diagnosis of various ophthalmological diseases. Efficient segmentation of the RVT with reduced runtime is essential for clinical purposes. Recently, convolutional neural networks (CNNs) have been used for RVT segmentation. However, these architectures typically apply fixed and standard size of convolution kernels for all blocks, which may be unsuitable for accurately capturing vessel scales. In addition, these kernels are applied using 3D convolution layers across all channel depths, leading to higher computational complexity. In this work, we propose a novel deep learning architecture. The main contribution consists of performing a convolution processing where kernel size is chosen with respect to vessel scale variation, in order to enhance the quality of the segmentation of vascular trees. In addition, the convolution processing is insured through several layers with 2D kernels, to reduce the computational complexity. The proposed architecture is evaluated on DRIVE database reaching an average accuracy and sensitivity respectively in the order of 97.69%and 91.69% in 0.75 second per fundus image.

The Retinal Vascular Tree (RVT) segmentation is required to diagnose various ocular pathologies.Recently, fundus images are acquired with higher resolution, which allows representing a large range of vessel thickness.However, standard Deep Learning (DL) architectures with static and small convolution size have failed to achieve higher segmentation performance.In this paper, we propose a novel DL architecture for RVT segmentation dedicated for high resolution fundus images.The idea consists at extending the U-net architecture by increasing (e.g.decreasing) convolution kernel size through convolution blocs, in correlation with downscale (e.g.upscale) of feature map dimensions.The proposed architecture is validated on HRF database, where average sensitivity is increased from 56% to 84%.

Accurate segmentation of blood vessels can make an important and effective contribution to the identification and diagnosis of ocular diseases such as diabetic retinopathy, glaucoma, and hypertension. Contrast enhancement is an essential component of any retinal blood vessel segmentation process. The consistency of contrast within an image will define the reliability of the segmentation. A new approach to dynamic segmentation of retinal blood vessels is proposed in this paper. Preprocessing, vessel segmentation, and post-processing are the three main stages of this method. The enhancement technique was integrated with dynamic preprocessing to improve segmentation performance. The DRIVE database was used to test the proposed method and analyze the results. The experimental results confirmed an improvement in segmentation where the use of dynamic processing increased the accuracy from 91.65% to 93.23%.

The Diabetic Retinopathy (DR) is a worldwide eye disease that causes visual damages and can leads to blindness. Therefore, the detection of the DR in the early stages is highly recommended. However, a delay is registered for ensuring early DR diagnosis which caused by the low-rate of the ophthalmologists, the deficiency of diagnosis equipment and the lack of mobility of elderly patients. In this paper, the main objective is to provide a mobile-aided screening system of moderate DR. Within this aim, we propose a classifier-based method which is based on detecting the Hard Exudate (HE) lesions that occur in moderate DR stage. A set of features are extracted to ensure an accurate and robust detection with respect to modest quality of fundus images. Moreover, the detection is provided in a low complexity processing to be suitable for mobile device. The aimed system corresponds to the implementation of the method on a smartphone associated to an optical lens for capturing fundus image. The system reached satisfactory screening performance where an accuracy of 98.36%, a sensitivity of 100% and specificity of 96.45% are registered using the DIARETDB1 fundus image databases. Moreover, the screening is performed in an average execution time of 2.68 seconds.

Nested loops present the most critical sections in several embedded real-time applications. To attain a higher performance, several optimization techniques are employed in order to increase parallelism. However, due to the tight requirements, they are either unable to achieve any execution time constraint or achieve it with a high code size, which presents an implementation limiting factor. In this paper, we propose a novel optimization approach that combines two techniques which are the delayed multidimensional retiming and the loop striping. It explores the solution space, which is composed by all parallelism cases proposed by both techniques, in order to provide the implementation that achieves the execution time constraint while using a lower code size which has not been attained by the other two techniques. We propose the theory of combining both techniques and efficient algorithms of selecting the parallelism transformations. The experimental results show that our optimization approach achieves execution times where each technique can not. Moreover, even if they can, it presents average improvements on the code size of 35.21% compared to the delayed multidimensional retiming and 16.38% compared to the loop striping.

Multidimensional retiming is an efficient optimization approach that ensures increasing a parallelism level in order to optimize the execution time. Two existing techniques called incremental and chained multidimensional retiming are based on this approach, which aim at achieving a full parallelism on loop body in order to schedule applications with a minimum cycle period. However, the cycle number increases in terms of parallelism level which presents a limiting factor to respect the execution time constraint of real-time applications. In this paper, we show how the minimal cycle period is achieved in multidimensional applications without applying a full parallelism. We present the theory of a novel technique, called delayed multidimensional retiming. Firstly, two efficient algorithms are presented where the first one insures the extraction of timing and data dependency properties of the application and the second one selects the set of data path for retiming. Then, we propose theorems to deduce a retiming function for the selected paths. Finally, a third algorithm describing the optimization approach is introduced. The experimental results show that our technique improves execution times in comparison to existing techniques. It achieves average improvements on the execution time of 41.57% compared to the Incremental technique and 11.55% compared to the Chained technique.