Diabetic retinopathy (DR) is a fast-spreading disease across the globe, which is caused by diabetes. The DR may lead the diabetic patients to complete vision loss. In this scenario, early identification of DR is more essential to recover the eyesight and provide help for timely treatment. The detection of DR can be manually performed by ophthalmologists and can also be done by an automated system. In the manual system, analysis and explanation of retinal fundus images need ophthalmologists, which is a time-consuming and very expensive task, but in the automated system, artificial intelligence is used to perform an imperative role in the area of ophthalmology and specifically in the early detection of diabetic retinopathy over the traditional detection approaches. Recently, numerous advanced studies related to the identification of DR have been reported. This paper presents a detailed review of the detection of DR with three major aspects; retinal datasets, DR detection methods, and performance evaluation metrics. Furthermore, this study also covers the author's observations and provides future directions in the field of diabetic retinopathy to overcome the research challenges for the research community.
In diabetic retinopathy (DR), the early signs that may lead the eyesight towards complete vision loss are considered as microaneurysms (MAs). The shape of these MAs is almost circular, and they have a darkish color and are tiny in size, which means they may be missed by manual analysis of ophthalmologists. In this case, accurate early detection of microaneurysms is helpful to cure DR before non-reversible blindness. In the proposed method, early detection of MAs is performed using a hybrid feature embedding approach of pre-trained CNN models, named as VGG-19 and Inception-v3. The performance of the proposed approach was evaluated using publicly available datasets, namely “E-Ophtha” and “DIARETDB1”, and achieved 96% and 94% classification accuracy, respectively. Furthermore, the developed approach outperformed the state-of-the-art approaches in terms of sensitivity and specificity for microaneurysms detection.
Brain hemorrhage is a serious and life-threatening condition. It can cause permanent and lifelong disability even when it is not fatal. The word hemorrhage denotes leakage of blood within the brain and this leakage of blood from capillaries causes stroke and adequate supply of oxygen to the brain is hindered. Modern imaging methods such as computed tomography (CT) and magnetic resonance imaging (MRI) are employed to get an idea regarding the extent of the damage. An early diagnosis and treatment can save lives and limit the adverse effects of a brain hemorrhage. In this case, a deep neural network (DNN) is an effective choice for the early identification and classification of brain hemorrhage for the timely recovery and treatment of an affected person. In this paper, the proposed research work is divided into two novel approaches, where, one for the classification and the other for volume calculation of brain hemorrhage. Two different datasets are used for two different techniques classification and volume. A novel algorithm is proposed to calculate the volume of hemorrhage using CT scan images. In the first approach, the ‘RSNA’ dataset is used to classify the brain hemorrhage types using transfer learning and achieved an accuracy of 93.77%. Furthermore, in the second approach, a novel algorithm has been proposed to calculate the volume of brain hemorrhage and achieved tremendous results as 1035.91 mm3 and 9.25 cm3, using the PhysioNet CT scan tomography dataset.
Existing image-to-image translation methods usually adopt an encoder-decoder structure to generate images. The encoder extracts the features of input images using a sequence of convolution layers until a bottleneck, and then, the intermediate features are decoded to the target image. However, the existence of bottleneck layer in such structure may lead to blurry and bad quality of the translated images, since different domain translations may be related to the global or local region in the input image or even in an abstract level. To prevent these problems, we propose the channel attention networks for image translation in this paper. It is a novel model that supports the multi-domain image-to-image translation using one single model. Conditioning on the target domain label, an auto-encoder-like network with multiple attention connections is trained to translate the input image into the target domain. The attention connections better shuttle the low-level information in the encoder to the decoder, which helps to preserve the structure. A multi-level attention mechanism is also designed in the proposed model to further improve the performance of our model. More specially, the feature maps in the encoder are first squeezed by average pooling and used to output a channel-wise attention mask. The attention mask softly determines which channels of the feature maps are translated and which channels are kept. By enforcing the model to learn a cyclic domain transformation during training, our model does not require paired training data, which greatly improves the versatility to different kinds of data. We experimentally demonstrated the effectiveness of our proposed model on the facial and clothing image translation tasks. The extensive ablations are also conducted to further validate the contribution of the proposed attention module used in our model.
12 June 2025 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Deep learning (DL) requires large amounts of labeled data, which is extremely time-consuming and laborintensive to obtain for medical image segmentation tasks. Metalearning focuses on developing learning strategies that enable quick adaptation to new tasks with limited labeled data. However, rich-class medical image segmentation datasets for constructing meta-learning multi-tasks are currently unavailable. In addition, data collected from various healthcare sites and devices may present significant distribution differences, potentially degrading model's performance. In this paper, we propose a task augmentation-based meta-learning method for retinal image segmentation (TAMS) to meet labor-intensive annotation demand. A retinal Lesion Simulation Algorithm (LSA) is proposed to automatically generate multi-class retinal disease datasets with pixel-level segmentation labels, such that metalearning tasks can be augmented without collecting data from various sources. In addition, a novel simulation function library is designed to control generation process and ensure interpretability. Moreover, a generative simulation network (GSNet) with an improved adversarial training strategy is introduced to maintain high-quality representations of complex retinal diseases. TAMS is evaluated on three different OCT and CFP image datasets, and comprehensive experiments have demonstrated that TAMS achieves superior segmentation performance than state-of-the-art models.