Showing papers by "Goutam Saha published in 2009"

In search of an optimization technique for Artificial Neural Network to classify abnormal heart sounds

Abstract: Artificial Neural Network (ANN) finds use in classification of heart sounds for its discriminative training ability and easy implementation. The selection of number of nodes for an ANN remains an important issue as an overparameterized ANN gets trained along with the redundant information that results in poor validation. Also a larger network means more computational expense, resulting more hardware and time related cost. Therefore, a compact and optimum design of neural network is needed towards real-time detection of pathological patterns, if any from heart sound signals. This work attempts to (i) design a compact form of output layer with less number of nodes than output classes, (ii) select a set of input features that are effective for identification of heart sound signals using Singular Value Decomposition (SVD), QR factorization with column pivoting (QRcp) and Fisher's F-ratio, (iii) make certain optimum selection of nodes in the hidden layer for a more effective ANN structure using SVD and (iv) select and prune weights based on the concept of local relative sensitivity index (LRSI) for empirically chosen overparameterized ANN structure for phonocardiogram (PCG) classification. It is observed that the proposed techniques perform better in terms of reduction of model residues and time complexity for classifying 12 different pathological cases and normal heart sound.

Investigating long-range correlation properties in EEG during complex cognitive tasks

Abstract: Previous work shows the presence of scale invariance and long-range correlations in ongoing and spontaneous activity of large scale brain responses (i.e. EEG), and such scaling behavior can also be modulated by simple sensory stimulus. However, little is known whether such alteration but not destruction in scaling properties also occurs during complex cognitive processing and if neuroplasticity plays any role in mediating such changes. In this study, we addressed these issues by investigating scaling properties of multivariate EEG signals obtained from two broad groups – artists and non-artists – while they performed complex tasks of perception and mental imagery of visual art objects. We found that brain regions showing increased correlation properties from rest were similar for both tasks, suggesting that brain networks responsible for visual perception are reactivated for mental imagery. Further, we observed that the two groups could be differentiated by scaling exponents and an artificial neural network based classifier achieved a classification efficiency of over 80%. These results altogether suggest that specific complex cognitive task demands and task-specific expertise can modify the temporal scale-free dynamics of brain responses.

Investigating neuromagnetic brain responses against chromatic flickering stimuli by wavelet entropies

Abstract: BACKGROUND: Photosensitive epilepsy is a type of reflexive epilepsy triggered by various visual stimuli including colourful ones. Despite the ubiquitous presence of colorful displays, brain responses against different colour combinations are not properly studied. METHODOLOGY/PRINCIPAL FINDINGS: Here, we studied the photosensitivity of the human brain against three types of chromatic flickering stimuli by recording neuromagnetic brain responses (magnetoencephalogram, MEG) from nine adult controls, an unmedicated patient, a medicated patient, and two controls age-matched with patients. Dynamical complexities of MEG signals were investigated by a family of wavelet entropies. Wavelet entropy is a newly proposed measure to characterize large scale brain responses, which quantifies the degree of order/disorder associated with a multi-frequency signal response. In particular, we found that as compared to the unmedicated patient, controls showed significantly larger wavelet entropy values. We also found that Renyi entropy is the most powerful feature for the participant classification. Finally, we also demonstrated the effect of combinational chromatic sensitivity on the underlying order/disorder in MEG signals. CONCLUSIONS/SIGNIFICANCE: Our results suggest that when perturbed by potentially epileptic-triggering stimulus, healthy human brain manages to maintain a non-deterministic, possibly nonlinear state, with high degree of disorder, but an epileptic brain represents a highly ordered state which making it prone to hyper-excitation. Further, certain colour combination was found to be more threatening than other combinations.