The world faces difficulties in terms of eye care, including treatment, quality of prevention, vision rehabilitation services, and scarcity of trained eye care experts. Early detection and diagnosis of ocular pathologies would enable forestall of visual impairment. One challenge that limits the adoption of computer-aided diagnosis tool by ophthalmologists is the number of sight-threatening rare pathologies, such as central retinal artery occlusion or anterior ischemic optic neuropathy, and others are usually ignored. In the past two decades, many publicly available datasets of color fundus images have been collected with a primary focus on diabetic retinopathy, glaucoma, age-related macular degeneration and few other frequent pathologies. To enable development of methods for automatic ocular disease classification of frequent diseases along with the rare pathologies, we have created a new Retinal Fundus Multi-disease Image Dataset (RFMiD). It consists of 3200 fundus images captured using three different fundus cameras with 46 conditions annotated through adjudicated consensus of two senior retinal experts. To the best of our knowledge, our dataset, RFMiD, is the only publicly available dataset that constitutes such a wide variety of diseases that appear in routine clinical settings. This dataset will enable the development of generalizable models for retinal screening.
18 September 2020 in Journal of Medical and Biological Engineering
Knowledge of muscle activity during movements is essential for understanding control strategies of human neuromuscular system. Human face is considered as source of information for revealing a person's affective state. This paper presents a method of recognizing two facial emotions from zygomaticus and corrugator muscles using two channel data acquisition system. Zygomaticus muscles aid in articulation of mouth, nose and cheeks whereas corrugator muscles draws the eyebrows downward producing vertical wrinkles on forehead. The bioelectric signals are recorded using bipolar surface electrode. This raw signal is filtered and windowed. The time domain features like root mean square value (RMS), median value and mean absolute value are extracted. Depending on features extracted thresholding is done and LabVIEW is used in messaging the status of nerve conduction. Thus in this paper we have proposed an optimize technique to record facial muscle potential and transmit the related information of the signal using LabVIEW and GSM (Global system for mobile communication) module using SMS (Short Messaging Service) to infer subject's mood state.
Response time between initial stage of diabetic macular edema (DME) and its severe stage leading to a higher risk of vision impairment is around 3 to 4 years. The paper presents a systematic computer aided model for detection of exudates and automatic screening of healthy and diseased fundus image for prompt management and early detection of symptoms to prevent the onset of DME. In DME macula is affected because of the deposits of exudates from leaky blood vessels causing blurred vision. The fundus images undergo three steps: pre-processing, feature extraction and classification. In the pre-processing stage, the images go through contrast limited adaptive histogram equalization, alternate sequential filtering, and thresholding for extraction of the blood vessel. Abnormalities in the form of exudates are found by comparing the vessel mask and Canny’s edge detected images. Feature selection is done from a set of features extracted to reduce the dimensionality and improve the accuracy of the model. The best-suited machine learning algorithm found is an Artificial Neural Network to train the features with an accuracy of 97.18% to prevent unnecessary vision loss.
The aim of this paper is to apply Imaging science to visualize foreign bodies’ like bright lesions, exudates and characterize them with a view to using them productively to examine and diagnose the severity of Diabetic Macular edema (DME) in human eyes. For the recognition of DME, a systemic flow is proposed. The methodology consists of using Contrast-Limiting Adaptive Histogram Equalization (CLAHE) and Alternate sequential filtering (AFS) techniques to obtain blood vessel mesh. The anatomical structures like optic disc are masked and the macular region is found for severity grading. Thresholding is applied to preserve the necessary details from the grayscale image. This image is processed with the edge detected image to obtain the exudates’ presence. Finally, the severity grading is done on retinal images to obtain bright lesions. The accuracy of the algorithm was found to be 89.9 percent.