This paper presents a new technique for automatically detecting and characterizing major brain lesions for diffusion-weighted imaging. The analytical framework consists of pre-processing, segmentation, features extraction and classification. For segmentation process, Fuzzy C-Means integrated with correlation template are proposed to detect the lesion region. The algorithm performance is evaluated using Jaccard and both false positive and false negative rates. Next, the features from wavelet transform are extracted from the region and fed into the rule-based classifier. Results demonstrated that FCM with correlation template offered the best performance for acute stroke segmentation with the highest rate of 0.77 Jaccard index. The classification accuracy for acute stroke, tumor, chronic stroke and necrosis are 94%, 97, 63% and 60%. In conclusion, the proposed hybrid analysis has the potential to be explored as a computer-aided tool to detect and diagnose of human brain lesion.
The research and development of the cochlear biomodels have initially started over a century ago. Since then, various types of approach have been implemented in trials to perfectly replicate the nature of the human auditory system. The evolution started with the implementation of mechanical elements into the cochlear biomodel operating in air and fluidic surrounding. However, due to the huge size of the mechanical cochlear biomodel, the microelectromechanical systems (MEMS) has been implemented in order to attain a life-sized cochlear biomodel. Researchers have looked into the possibilities of fabricating the MEMS cochlear biomodel in air and fluidic mediums. In this paper, the mechanical and MEMS cochlear biomodel implementations will be reviewed. The key part in modelling the cochlea for human auditory system is to mimic closely its nature and capabilities in terms of the geometrical design, material properties and sensory performance.
Electroencephalogram (EEG) based classification has achieved a promising performance using deep learning models like Convolutional Neural Network. Various pre-processing strategies such as smoothing the EEG data or filtering are commonly used to pre-process the captured EEG signal before the subsequent feature extraction and classification while hyperparameters tuning might help to improve the classification performance. As well, the number of layers used in the CNN can affect the performance of the classification. In this paper, the number of layers needed for the CNN to classify the EEG data correctly, the effect of apply smoothing to pre-process the EEG signal for modern end-to-end CNN and the effect of enabling hyperparameters tuning during the training phase of CNN is investigated and analyzed. Two CNN models, namely Deep CNN with 5 layers and Shallow CNN with 1 layer, with convincing classification accuracy on motor execution classification as reported in the literature were chosen for this study. Both the CNN models are trained on EEG motor execution dataset with different training strategies and dataset pre-processing. Based on the obtained training and test classification accuracy, Shallow CNN trained with enabling hyper parameters tuning and without smoothing the EEG data achieved the best classification accuracy with average training accuracy of 99.9% and test accuracy of 96.87%. This indicates that CNN does not need to have many layers to correctly classify the motor execution data and the EEG data does not require smoothing.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.