Search

Analysis of data for identifying patterns and building models has been used as a strong tool in different domains, including medical domains. In this paper, we analyse the registry of brain stroke patients collected over fifteen years in south London hospitals, known as South London Stroke Register. Our attempt is to identify the similar patterns between patients’ background and living conditions, their cognitive ability, the treatments they received, and the speed of their cognitive recovery; based on which most effective treatment can be predicted for new admitted patients. We designed a novel strategy which takes into account two different approaches. First is to predict, for each of the potential intervention treatments, whether that particular treatment would lead to recovery of a new patient or not. Second is to suggest a treatment (treatments) for the patient based on those that were given to the patients who have recovered and are most similar to the new patient. We built different classifiers using various state of the art machine learning algorithms. These algorithms were evaluated and compared based on three performance metrics, defined in this paper. Given that time is very crucial for stroke patients, main motivation of this research work is identifying the most effective treatment immediately for a new patient, and potentially increase the probability of their cognitive recovery.

In this paper, a method for automatic recognition of acute stroke model using non-contrast computed tomography brain images is presented. The complexity of the task lies in the fact that the dataset consists of a very small number of images. To solve the problem, we used the traditional computer vision methods and a convolutional neural network consisting of a segmentator and classifier. To increase the dataset, augmentations and sub images were used. Experiments with real CT images using validation and test samples showed that even on an extremely small dataset it is possible to train a model that will successfully cope with the classification and segmentation of images. We also proposed a way to increase the interpretability of the model.

Objective: Timely and effective discrimination of hemorrhagic stroke and ischemic stroke can significantly improve the prognosis. Current discrimination is expensive and has the disadvantage of having to be in contact with the patient. Based on animal experiments, in this paper, microwave measurement technique is used to study the discrimination of two stroke types. Method: In the experiments, 10 rabbits (5 cerebral hemorrhage and 5 cerebral ischemia) are selected. Cerebral hemorrhage is induced by injecting autologous blood (1 to 4 mL) into the brain of rabbits, and the cerebral ischemia is induced by bilateral common carotid artery ligation and femoral artery blood extraction. The two groups are monitored by a 16-channel microwave detection system to obtain the reflection parameter caused by pathological changes in the brain. After redundancy removed from original data, support vector machine (SVM) is used to identify the type and severity of two types of stroke. Findings: The study shows that the microwave-based stroke identification system can effectively distinguish the cerebral hemorrhage model and the cerebral ischemia model. The experimental system is very promising in pre-hospital stroke type identification because of low cost, non-invasive, simple operation and rapid measurement.

Public health is one of the most concerns at the worldwide. Brain ischemic stroke is an acute cerebral vascular disease, which is likely to cause long-term disabilities and death. It occurs in most stroke patients. Brain Magnetic Resonance Imaging (MRI) is one of the essential non-invasive modalities that provide a contrast imaging to visualize and detections lesions. Brain ischemic stroke segmentation in MRI has attracted the attention of medical doctors and researches since variations in structural and contrast of medical data. Several proposals have been designed throughout the years comprising a different strategy of brain segmentation. In particular, in this paper we analyse a segmentation methods used for detection and localization brain ischemic stroke. That the goal has been presented to differentiate between the lesions with the normal region. The Spatial Fuzzy C Means (SFCM) and methods based regions are developed in order to obtain a robust, rapid, efficient, precious and precocious detection of acute stroke lesion from images data issues by MRI with diffusion-weighted imaging (DWI) and perfusion-weighted imaging (PWI). The validation purpose was performed by comparing resulting segmentation to the manual contours traced by an expert. Results show that the SFCM appeared efficient in detection of acute with a accuracy value of 99.1% in PWI-MTT and of 47.44% in DWI and an timing average in order to one second. However, the accuracy rate of regions growing in order to 17.40% in DWI and to 71.30% in PWI.

Detection of acute stroke signs in non-contrast CT images is a challenging task. The intensity and texture variations in pathological regions are subtle and can be confounded by normal physiological changes or by old lesions. In this paper we investigate the use of contextual information for stroke sign detection. In particular, the appearance of the contralateral anatomy and the atlas-encoded spatial location are incorporated into a Convolutional Neural Network (CNN) architecture. CNNs are trained separately for the detection of dense vessels and of ischaemia. The network performance is evaluated on 170 datasets by cross-validation. We find that atlas location is important for dense vessel detection, but is less useful for ischaemia, whereas bilateral comparison is crucial for detection of ischaemia.