Ramakrishnan Swaminathan

In this study, an attempt has been made to differentiate Novel Coronavirus-2019 (COVID-19) conditions from healthy subjects in Chest radiographs using a simplified end-to-end Convolutional Neural Network (CNN) model and occlusion sensitivity maps. Early detection and faster automated screening of the COVID-19 patients is essential. For this, the images are considered from publicly available datasets. Significant biomarkers representing critical image features are extracted from CNN by experimentally investigating on cross-validation methods and hyperparameter settings. The performance of the network is evaluated using standard metrics. Perturbation based occlusion sensitivity maps are employed on the features obtained from the classification model to visualise the localization of abnormal areas. Results demonstrate that the simplified CNN model with optimised parameters is able to extract significant features with a sensitivity of 97.35% and F-measure of 96.71% to detect COVID-19 images. The algorithm achieves an Area Under the Curve-Receiver Operating Characteristic score of 99.4% with Matthews correlation coefficient of 0.93. High value of Diagnostic odds ratio is also obtained. Occlusion sensitivity maps provide precise localization of abnormal regions by identifying COVID-19 conditions. As early detection through chest radiographic images are useful for automated screening of the disease, this method appears to be clinically relevant in providing a visual diagnostic solution using a simplified and efficient model.

Background and Objective: Computer aided diagnostics of Pulmonary Tuberculosis in chest radiographs relies on the differentiation of subtle and non-specific alterations in the images. In this study, an attempt has been made to identify and classify Tuberculosis conditions from healthy subjects in chest radiographs using integrated local feature descriptors and variants of extreme learning machine. Methods: Lung fields in the chest images are segmented using Reaction Diffusion Level Set method. Local feature descriptors such as Median Robust Extended Local Binary Patterns and Gradient Local Ternary Patterns are extracted. Extreme Learning Machine (ELM) and Online Sequential ELM (OSELM) classifiers are employed to identify Tuberculosis conditions and, their performances are analysed using standard metrics. Results: Results show that the adopted segmentation method is able to delineate lung fields in both healthy and Tuberculosis images. Extracted features are statistically significant even in images with inter and intra subject variability. Sigmoid activation function yields accuracy and sensitivity values greater than 98% for both the classifiers. Highest sensitivity is observed with OSELM for minimal significant features in detecting Tuberculosis images. Conclusion: As ELM based method is able to differentiate the subtle changes in inter and intra subject variations of chest X-ray images, the proposed methodology seems to be useful for computer-based detection of Pulmonary Tuberculosis.

In this work, an attempt has been made to analyze the influence of the frequencies bands in uterine electromyography (uEMG) signals on the detection of preterm birth. The signals recorded from the women's abdomen during pregnancy are considered in this study. The signals are subjected to preprocessing using digital bandpass Butterworth filter and decomposed into different frequency bands namely, 0.3-1.0 Hz (F1), 1.0-2.0 Hz (F2) and 2.0-3.0Hz (F3). Spectral features namely, peak magnitude, peak frequency, mean frequency and median frequency are extracted from the power spectrum. Classification models namely, k-nearest neighbor, support vector machine and random forest are employed to distinguish the term and preterm conditions. The results show that the features extracted from these frequency bands are able to differentiate term and preterm condition. Particularly, the frequency band F3 performs better than other frequency bands. The features associated with these frequencies along with random forest classification model achieves a maximum accuracy of 75.2%. Thus, these measures could be used to accurately detect the preterm birth well in advance. © 2021 European Federation for Medical Informatics (EFMI) and IOS Press.

Viscoelastic properties of skeletal muscle tissue are known to be impacted by fatiguing contractions. In this study, an attempt has been made to utilize myotonometry for analyzing the relationship between muscle viscoelasticity and contractile behaviors in a fatiguing task. For this purpose, thirteen young healthy volunteers are recruited to perform the fatiguing isometric task and the time to task failure (TTF) is recorded. Myotonometric parameters and simultaneous surface electromyographic (sEMG) signals are recorded from the Biceps Brachii muscle of the flexed arm. The correlation between myotonometric parameters and TTF is further analyzed. Cross-validation with sEMG features is also performed. Stiffness of muscle has a positive correlation with TTF in the left hand (p<0.05). Damping property of the nonfatigued muscle is positively associated with the fatigue-induced changes in amplitude features of sEMG signal in the right hand (p<0.05). The normalized rate of change of mean frequency of sEMG signal has a positive correlation with stiffness values in both of the hands (p<0.05). Muscle viscoelasticity is demonstrated to influence the progression of fatigue, although the difference in motor control due to handedness is also found to be an important factor. The results are promising to improve the understanding of the effect of muscle mechanics in fatigue-induced task failure.

Analysis of uterine contractions using electromyography signals is gaining importance due to its capability to measure the dynamics of uterus. Uterine electromyography (uEMG) provides information on the nature of uterine contractions non-invasively. In this study, the fluctuations in uEMG signals associated with Term pregnancies are analyzed. For this, Term uEMG signals corresponding to second (T1) and third (T2) trimesters are considered. The signals are subjected to Adaptive Fractal Analysis (AFA), wherein a global trend is obtained by using overlapping windows of three orders namely, 25%, 50% and 75%. The signals are detrended and the fluctuation function is estimated. Two Hurst exponent features computed at short range (Hs) and long range (Hl) are extracted and statistically analyzed. Results show that AFA is able to characterize variations in the fluctuations of Term delivery signals. The feature values are observed to vary significantly during different weeks of gestation. It is found that features of T2 signals are higher than that of T1 signals for all the considered overlaps, indicating that T2 signals possess smoother characteristics than T1 signals. Further, coefficient of variation is observed to be low, indicating that these features are able to handle the inter-subject variations in Term signals. Therefore, it appears that the proposed approach could aid in investigation of progressive changes in uterine contractions during Term pregnancies.

In this work, an attempt has been made to utilise the time-frequency spectrum obtained using generalised Warblet transform (GWT) for fatigue analysis. Signals are acquired from the biceps brachii muscles of 20 healthy volunteers during isometric contractions. The first and last 500 ms lengths of a signal are assumed as non-fatigue and fatigue zones respectively. Further, the signals from these zones are subjected to GWT for the computation of time-frequency spectrum. Features such as instantaneous mean frequency (IMNF), instantaneous median frequency (IMDF), instantaneous spectral entropy (ISPEn), and instantaneous spectral skewness (ISSkw) are estimated. The results show that the IMNF, IMDF and ISPEn increased by 24%, 34% and 36% respectively in non-fatigue condition. In contrast, 22% higher ISSkw is observed for fatigue condition. The statistical analysis indicates that the features are significant with p < 0.001. It appears that the current method is useful in analysing muscle fatigue disorders using sEMG signals.

Diabetes mellitus is a group of metabolic disease, which has become globally prevalent, and affects a large population in socio-economically backward countries in Asian continent. Chronic diabetes can lead to ulceration in the plantar region and may result in amputation. Assessment of mechanical properties of plantar tissues can aid in early diagnosis of ulceration. Myotonometry, a technique to measure dynamic stiffness, is preferred due to its noninvasiveness, easy employability, and rapid investigation. In this study, an attempt has been made to analyze the changes in biomechanical properties of plantar soft tissue in diabetes. MyotonPro, a handheld device, is used for this purpose. 43 diabetic subjects with varied duration of diabetes are recruited. Site-specific mechanical properties of the plantar region for both the feet are acquired and statistical analysis is performed. Results show that the MyotonPro is able to differentiate the stages of diabetes. It is seen that there is a spatial variability in the mechanical properties of the plantar. Additionally, it is observed that there is a significant increment in the plantar stiffness value in the group with higher diabetic age (p < 0.05). Further, significant changes in dynamic mechanical properties are also observed in submetatarsal region. Additionally, a right-left asymmetry has been observed in frequency and stiffness values for later stages of diabetes. This study demonstrated the feasibility of MyotonPro in discriminating the stages of diabetic period. Thus, the proposed approach could be useful in early diagnosis of foot ulceration for various clinical conditions.

In this work, automated abnormality detection using keypoint information from Speeded-Up Robust feature (SURF) and Scale Invariant Feature Transform (SIFT) descriptors in chest Radiographic (CR) images is investigated and compared. Computerized image analysis using artificial intelligence is crucial to detect subtle and non-specific alterations of Tuberculosis (TB). For this, the healthy and TB CRs are subjected to lung field segmentation. SURF and SIFT keypoints are extracted from the segmented lung images. Statistical features from keypoints, its scale and orientation are computed. Discrimination of TB from healthy is performed using SVM. Results show that the SURF and SIFT methods are able to extract local keypoint information in CRs. Linear SVM is found to perform better with precision of 88.9% and AUC of 91% in TB detection for combined features. Hence, the application of keypoint techniques is found to have clinical relevance in the automated screening of non-specific TB abnormalities using CRs. © 2021 European Federation for Medical Informatics (EFMI) and IOS Press.

Alzheimer's Disease (AD) is a progressive fatal neurodegenerative disorder that causes cognitive decline in affected people. Image processing of brain MR images can aid in identifying significant imaging biomarkers for detection of AD and its prodromal stage Mild Cognitive Impairment (MCI). Bidimensional multiscale entropy-based texture analysis is a new approach to quantify the textural variations in images at multiple scales. This work is based on the application of bidimensional multiscale entropy for analyzing AD induced textural alterations in lateral ventricles of the brain MR images. For this T1 weighted MR brain images of normal, MCI and AD subjects are obtained from public database. Lateral ventricles (LV) are delineated using reaction–diffusion level set technique from transaxial image slice with high accuracy. Bidimensional multiscale entropy is then applied on segmented LV to extract entropy features at multiple image scales and complexity indices are evaluated for each scale to study textural variations. The parameters such as tolerance factor, window lengths and scales for computation of multiscale entropy for significant differentiation amongst the healthy and diseased subjects are experimentally evaluated. The obtained entropy values from healthy subjects are observed to be significantly lower from the pathological subjects across scales. Classification with extracted features using a linear discriminant classifier achieves an accuracy of 80.1% and 87.6% for Normal vs MCI and Normal vs AD classes, respectively. The proposed multiscale entropy-based approach captures the textural alterations in lateral ventricles of brain MR images and furthermore, can be used as automated tool for early diagnosis of AD.

In this work, an attempt has been made to classify various emotional states in Electrodermal Activity (EDA) signals using modified Hjorth features and non-parametric classifiers. For this, the EDA signals are collected from a publicly available online database. The EDA is decomposed into SCL (Skin Conductance Level) and SCR (Skin Conductance Response). Five features, namely activity, mobility, complexity, chaos, and hazard, collectively known as modified Hjorth features, are extracted from SCR and SCL. Four non-parametric classifiers, namely, random forest, k-nearest neighbor, support vector machine, and rotation forest, are used for the classification. The results demonstrate that the proposed approach can classify the emotional states in EDA. Most of the features exhibit statistical significance in discriminating emotional states. It is found that the combination of modified Hjorth features and rotation forest is most accurate in classifying the emotional states. Thus, the result demonstrates that this method can recognize valence and arousal dimensions under various clinical conditions. © 2021 European Federation for Medical Informatics (EFMI) and IOS Press.