Bi-modality image classification based on independent component analysis / 生物医学工程学杂志

We in the present research proposed a classification method that applied infomax independent component analysis (ICA) to respectively extract single modality features of structural magnetic resonance imaging (sMRI) and positron emission tomography (PET). And then we combined these two features by using a method of weight combination. We found that the present method was able to improve the accurate diagnosis of Alzheimer's disease (AD) and mild cognitive impairment (MCI). Compared AD to healthy controls (HC): the study achieved a classification accuracy of 93.75%, with a sensitivity of 100% and a specificity of 87.64%. Compared MCI to HC: classification accuracy was 89.35%, with a sensitivity of 81.85% and a specificity of 99.36%. The experimental results showed that the bi-modality method performed better than the individual modality in comparison to classification accuracy.

Research on a successively increasing feature selection algorithm of EEG signal for driving fatigue based on SVM / 生物医学工程学杂志

Electroencephalogram (EEG) signals provide an objective physiological index for the identification of the driver's fatigue state. It is very important to choose appropriate channels and EEG signal features adaptively due to the features varying with different subjects and time. A support vector machine (SVM) based increasing feature selection algorithm for driving fatigue EEG classification is presented in this paper. The algorithm is a method to select EEG channels and features for driving fatigue adaptively in an ascending order. We can select the optimal feature each time from the remaining candidate features using the optimized SVM model minimum error rate as the index. The experimental calculation has characteristics of using 16 electrode channels which cover the whole head in the main area, of selecting 208 candidate features as the initial set, of selecting to the EEG data calculation recorded in 5 different time periods of a subject, and of choosing error rate of 2% as the algorithm termination condition. The selected features and models, therefore, can reach a high level of classification and generalization ability.