Inter-subject and inter-session variabilities pose a significant challenge in Electroencephalogram (EEG) based Brain Computer Interface (BCI) systems. Furthermore, high dimensional EEG montages introduce huge computational burden due to excessive number of channels involved. Two experimental, i.e., inter-session and inter-subject, variabilities of EEG dynamics during motor imagery (MI) tasks are investigated in this paper. In particular, the effect on the performance of the BCIs due to day-to-day variability in EEG dynamics during the alterations in cognitive stages is explored. In addition, the inter-subject BCIs feasibility between cortically synchronized and desynchronized subject pairs on pairwise performance associativity (PPA) is further examined. Moreover, the consequences to these two contexts by the integration of spatial brain dynamics of different number of channels from specific regions of brain are discussed. The proposed approach is validated on real BCI dataset containing EEG data from four classes of MI tasks, i.e., Left/Right-Hand, both Feet and Tongue, subjected prior to a preprocessing of three different spatial filtering techniques. Experimental results have shown that a maximum classification accuracy of around 58% was achieved for the inter-subject experimental case, whereas a 31% deviation was noticed in the classification accuracies across two sessions during the inter-session experimental case. In conclusion, BCIs, without the subject-and session-specific calibration and with lesser number of channels employed, play a vital role while promoting a generic and efficient framework for plug and play use.

We present a framework for handwritten Bangla digit recognition using Sparse Representation Classifier. The classifier assumes that a test sample can be represented as a linear combination of the train samples from its native class. Hence, a test sample can be represented using a dictionary constructed from the train samples. The most sparse linear representation of the test sample in terms of this dictionary can be efficiently computed through ℓ ₁ -minimization, and can be exploited to classify the test sample. We applied Sparse Representation Classifier on the image zone density, an image domain statistical feature extracted from the character image, to classify the Bangla numerals. This is a novel approach for Bangla Optical Character Recognition, and demonstrates an excellent accuracy of 94% on the off-line handwritten Bangla numeral database CMATERdb 3.1.1. This result is promising, and should be investigated further.

Cardiovascular autonomic neuropathy (CAN) is one of the most overlooked complications associated with diabetes. It is characterized by damage in the autonomic nerves regulating heart rate and vascular compliance. Ewing battery is currently the diagnostic tool of choice but is unable to detect sub-clinical CAN and requires patient cooperation. In addition, appropriate timing (day/night) of CAN diagnostic test was not explored in the past. Therefore, a novel approach is proposed herein to investigate the feasibility of using heart rate variability (HRV) features over 24 hours embedded within machine learning algorithms to provide a complete screening for patients suffering from CAN. 24-hour Holter ECG data were acquired from a Bangladeshi cohort (n = 95 patients [75 Diabetic and 25 healthy]). HRV features were extracted from every 5-minute segment of the HRV signal and used as input to four machine learning algorithms for hourly training and testing. A complete hierarchical step by step diagnosis procedure (4 tests) was developed; namely test 1 to check for being healthy or diabetic; test 2 to check for any microvascular complications (including neuropathy such as CAN, peripheral neuropathy (DPN), nephropathy (NEP), and retinopathy (RET)) or not; test 3 to check for presence of only CAN; test 4 to check for combined or multiple complications along with CAN. The highest levels of performance were achieved with accuracy measures of 85.5% (test 1 - convolutional neural network (CNN)), 98.5% (test 2 - CNN), 98.3% (test 3 - one-class support vector machines (SVM)), and 90.9% (test 4 - random forest). Hours 7:00 AM and 7:00 PM were found to be most significant in the diagnosis of CAN in diabetic patients (test 1, 3, and 4). Early screening of CAN by our proposed models could help primary healthcare centers stratify the risk leading to early treatment in preventing sudden cardiac death due to silent myocardial infarction. The approach is considered to be simple and effective, especially for under-resourced clinical settings.

We propose event-related cortical sources estimation from subject-independent electroencephalography (EEG) recordings for motor imagery brain computer interface (BCI). By using wavelet-based maximum entropy on the mean (wMEM), task-specific EEG channels are selected to predict right hand and right foot sensorimotor tasks, employing common spatial pattern (CSP) and regularized common spatial pattern (RCSP). EEG from five healthy individuals (Dataset IVa, BCI Competition III) were evaluated by a cross-subject paradigm. Prediction performance was evaluated via a two-layer feed-forward neural network, where the classifier was trained and tested by data from two subjects independently. On average, the overall mean prediction accuracies obtained using all 118 channels are (55.98±6.53) and (71.20±5.32) in cases of CSP and RCSP, respectively, which are slightly lower than the accuracies obtained using only the selected channels, i.e., (58.95±6.90) and (71.41±6.65), respectively. The highest mean prediction accuracy achieved for a specific subject pair by using selected EEG channels was on average (90.36±5.59) and outperformed that achieved by using all available channels (86.07 ± 10.71). Spatially projected cortical sources approximated using wMEM may be useful for capturing inter-subject associative sensorimotor brain dynamics and pave the way toward an enhanced subject-independent BCI.