According to the WHO, Cerebrovascular Stroke, or CS, is the second largest cause of death worldwide. Current diagnosis of CS relies on labor and cost intensive neuroimaging techniques, unsuitable for areas with inadequate access to quality medical facilities. Thus, there is a great need for an efficient diagnosis alternative. StrokeSave is a platform for users to self-diagnose for prevalence to stroke. The mobile app is continuously updated with heart rate, blood pressure, and blood oxygen data from sensors on the patient wrist. Once these measurements reach a threshold for possible stroke, the patient takes facial images and vocal recordings to screen for paralysis attributed to stroke. A custom designed lens attached to a phone's camera then takes retinal images for the deep learning model to classify based on presence of retinopathy and sends a comprehensive diagnosis. The deep learning model, which consists of a RNN trained on 100 voice slurred audio files, a SVM trained on 410 vascular data points, and a CNN trained on 520 retinopathy images, achieved a holistic accuracy of 95.0 percent when validated on 327 samples. This value exceeds that of clinical examination accuracy, which is around 40 to 89 percent, further demonstrating the vital utility of such a medical device. Through this automated platform, users receive efficient, highly accurate diagnosis without professional medical assistance, revolutionizing medical diagnosis of CS and potentially saving millions of lives.
Deep Dive into StrokeSave: A Novel, High-Performance Mobile Application for Stroke Diagnosis using Deep Learning and Computer Vision.
According to the WHO, Cerebrovascular Stroke, or CS, is the second largest cause of death worldwide. Current diagnosis of CS relies on labor and cost intensive neuroimaging techniques, unsuitable for areas with inadequate access to quality medical facilities. Thus, there is a great need for an efficient diagnosis alternative. StrokeSave is a platform for users to self-diagnose for prevalence to stroke. The mobile app is continuously updated with heart rate, blood pressure, and blood oxygen data from sensors on the patient wrist. Once these measurements reach a threshold for possible stroke, the patient takes facial images and vocal recordings to screen for paralysis attributed to stroke. A custom designed lens attached to a phone’s camera then takes retinal images for the deep learning model to classify based on presence of retinopathy and sends a comprehensive diagnosis. The deep learning model, which consists of a RNN trained on 100 voice slurred audio files, a SVM trained on 410 vas
According to the WHO, cerebrovascular stroke is the second largest cause of death worldwide, accounting for more than 5 million deaths in 2017 alone. Conventional detection methods involve physician administered exams including blood tests, angiograms, carotid ultrasounds, CT, and MRI scans. These neurological scans such as MRI are first of all costly, and are generally obtained 12 hours after the onset of symptoms, thus preventing patients from receiving treatment in the optimal timeframe. Thus, early, automated diagnosis methods using other biomarkers of stroke have been a central focus on cerebrovascular accident (CVA) research. Currently, there are no existing portable stroke screening platform available to the public. This is a dire issue requiring immediate action, since over 140,000 people die of stroke in the US and in every 40 seconds, someone is afflicted with stroke. Since current diagnosis methods yield accuracies up to 80% and stroke diagnosis is largely inaccessible in areas without access to quality medical facilities, a holistic patient centered early diagnosis application is critical to address this pressing issue. Moreover, recent advances in the applications of deep learning and computer vision, backend databases, and mobile application development allow us to engineer a timely, inexpensive, accurate stroke detection platform with far higher accuracy than current medical professionals.
Cerebrovascular accident (CVA) is a disease in which blood flow to a part of the brain is stopped either by a blockage or rupture of a blood vessel. Using diagnosis techniques such as blood tests, angiograms, carotid ultrasounds, CT scans, and echocardiograms, physicians determine the best course of treatment, which usually involves a combination of clot busters (tPA) and procedures such as thrombectomy.
Transient ischemic attacks (TIA) are temporary blockages in the brain often caused by a buildup of plaque in major arteries. Several symptoms of TIA, such as a sagging facial features, arm weakness, and speech difficulties, can be assessed to determine whether the patient has a stroke, and further steps can be taken to prevent a major stroke in the future. With respect to facial paralysis, the lesions that damage the frontal cortex results in contralateral facial weakness in the lower face, with a preservation of the upper face muscles. Additional factors in prediction of onset of an impending stroke include vascular data manifested in heart rate, blood pressure, and blood oxygen level measurements. Spikes in heart rate and blood pressure are associated with the sudden blockage of a blood vessel, as is a sudden drop in blood oxygen level. Furthermore, to predict the onset of an impending stroke, researchers at the American Atherosclerosis Society have uncovered several retinal and atrial biomarkers that can be attributed to an impending stroke, including hypertensive retinopathy. Hypertensive retinopathy lesions mainly result from small arteriosclerosis and uncontrolled hypertension, which lead to retinal ischemia and breakdown of blood retinal barrier. As the retina is the window to cerebral conditions, changes in the retina directly mirror primary cerebral changes characteristic of an impending stroke, such as the increased vessel permeability due to blood brain barrier breakdown. As such, assessing presence of retinopathy like symptoms is essential to assess a long term risk for stroke.
Machine Learning (ML) is a subset of Artificial Intelligence (AI) that involves giving automated systems the ability to “learn” data rather than being explicitly programmed.The Neural Network (NN) is a machine learning algorithm modeled after the brain composed of layers of ’neurons connected by ‘synapses’. As data is propagated through the NN, weights and biases are propounded, further allowing the Neural Network to learn data given, forming a ML model. Computer Vision (CV) is an interdisciplinary field within the field of computer science that deals with how computers can gain a high-level understanding of digital images or videos.
Learning algorithms together to learn data representations rather than task-specific algorithms.
Furthermore, combining Deep Learning-based models, such as CNNs, RNNs, and SVMs, combined with powerful CV algorithms such as the Active Appearance Model (AAM), yield exceptional accuracies.
In order to address to issue, our platform, StrokeSave, is built upon the fact that most areas have access to a wireless, mobile device, lens attachment, and medical grade device which can measure vascular data such as blood pressure, blood oxygen, and heart rate, all costing less than a combined $100. SAS is basically a robust android diagnosis application that incorporates a 3 tier diagnosis procedure for predicting the risk of the onset of stroke, ensuring that patients can receive treatment in an optimal timeframe. Our platform utilizes real-time vascular patient data along with Deep Learning and Com
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
