Abstract

Abstract
The creation of retinal mosaics from sets of fundus photographs can significantly reduce the time spent on the diabetic retinopathy (DR) screening, because through mosaic analysis the ophthalmologists can examine several portions of the eye at a single glance and, consequently, detect and grade DR more easily. Like most of the methods described in the literature, this methodology includes two main steps: image registration and image blending. In the registration step, relevant keypoints are detected on all images, the transformation matrices are estimated based on the correspondences between those keypoints and the images are reprojected into the same coordinate system. However, the main contributions of this work are in the blending step. In order to combine the overlapping images, a color compensation is applied to those images and a distance-based map of weights is computed for each one. The methodology is applied to two different datasets and the mosaics obtained for one of them are visually compared with the results of two state-of-the-art methods. The mosaics obtained with our method present good quality and they can be used for DR grading. © 2018, Springer International Publishing AG, part of Springer Nature.

Abstract
The classification of retinal vessels into arteries and veins in eye fundus images is a relevant task for the automatic assessment of vascular changes. This paper presents a new approach to solve this problem by means of a Fully-Connected Convolutional Neural Network that is specifically adapted for artery/vein classification. For this, a loss function that focuses only on pixels belonging to the retinal vessel tree is built. The relevance of providing the model with different chromatic components of the source images is also analyzed. The performance of the proposed method is evaluated on the RITE dataset of retinal images, achieving promising results, with an accuracy of 96 % on large caliber vessels, and an overall accuracy of 84 %. © 2018, Springer International Publishing AG, part of Springer Nature.

Abstract

Abstract
Image dehazing tries to solve an undesired loss of visibility in outdoor images due to the presence of fog. Recently, machine-learning techniques have shown great dehazing ability. However, in order to be trained, they require training sets with pairs of foggy images and their clean counterparts, or a depth-map. In this paper, we propose to learn the appearance of fog from weakly-labeled data. Specifically, we only require a single label per-image stating if it contains fog or not. Based on the Multiple-Instance Learning framework, we propose a model that can learn from image-level labels to predict if an image contains haze reasoning at a local level. Fog detection performance of the proposed method compares favorably with two popular techniques, and the attention maps generated by the model demonstrate that it effectively learns to disregard sky regions as indicative of the presence of fog, a common pitfall of current image dehazing techniques.

Abstract
This paper introduces a novel strategy for the task of simultaneously locating two key anatomical landmarks in retinal images of the eye fundus, namely the optic disc and the fovea. For that, instead of attempting to classify each pixel as belonging to the background, the optic disc, or the fovea center, which would lead to a highly class-imbalanced setting, the problem is reformulated as a pixelwise regression task. The regressed quantity consists of the distance from the closest landmark of interest. A Fully-Convolutional Deep Neural Network is optimized to predict this distance for each image location, implicitly casting the problem into a per-pixel Multi-Task Learning approach by which a globally consistent distribution of distances across the entire image can be learned. Once trained, the two minimal distances predicted by the model are selected as the locations of the optic disc and the fovea. The joint learning of every pixel position relative to the optic disc and the fovea favors an automatic understanding of the overall anatomical distribution. This results in an effective technique that can detect both locations simultaneously, as opposed to previous methods that handle both tasks separately. Comprehensive experimental results on a large public dataset validate the proposed approach.