Search

In this paper, a novel feature selection method is proposed for the categorization of electrocardiogram (ECG) heartbeats. The proposed technique uses the Fisher ratio and BAT optimization algorithm to obtain the best feature set for ECG classification. The MIT-BIH arrhythmia database contains sixteen classes of the ECG heartbeats. The MIT-BIH ECG arrhythmia database divided into intra-patient and inter-patient schemes to be used in this study. The proposed feature selection methodology works in following steps: firstly, features are extracted using empirical wavelet transform (EWT) and then higher-order statistics, as well as symbolic features, are computed for each decomposed mode of EWT. Thereafter, the complete feature vector is obtained by the conjunction of EWT based features and RR interval features. Secondly, for feature selection, the Fisher ratio is utilized. It is optimized by using BAT algorithm so as to have maximal discrimination of the between classes. Finally, in the classification step, the k-nearest neighbor classifier is used to classify the heartbeats. The performance measures i.e., accuracy, sensitivity, positive predictivity, specificity for intra-patient scheme are 99.80%, 99.80%, 99.80%, 99.987% and for inter-patient scheme are 97.59%, 97.589%, 97.589%, 99.196% respectively. The proposed feature selection technique outperforms the other state of art feature selection methods.

The most prevailing and most deadly form of the brain tumor is glioma, with a mortality rate of around 70%. Therefore, early and efficient treatment is necessary for improving patients' wellbeing. The MRI method has become a popular method for identifying these tumors, but the amount of data retrieved limits the use of quantitative measures that are accurate and reproducible. Therefore, Automating brain tumor segmentation from MR images is essential for tumor analysis and monitoring. For successful tumor delineation, accurate and reliable tumor segmentation techniques are used for gliomas, which are aggressive and heterogeneous. With the use of CNNs, which are widely utilized for biomedical image segmentation, the accuracy of the current state-of-the-art has greatly improved on the problem of segmenting brain tumors. This paper proposes a M U-Net architecture with a PReLU activation layer and smaller input kernels to increase the performance of segmenting tumor regions in MR images. An augmentation of data combined with normalization during preprocessing is applied to rectify the issue of class-imbalance in tumor segmentation. This experiment uses the BRATS 2017 dataset. It achieved accuracy, sensitivity, specificity of 97.94%, 99.9%, and 92.58% respectively.

In the past decade, there has been a remarkable evolution of convolutional neural networks (CNN) for biomedical image processing. These improvements are inculcated in the basic deep learning-based models for computer-aided detection and prognosis of various ailments. But implementation of these CNN based networks is highly dependent on large data in case of supervised learning processes. This is needed to tackle overfitting issues which is a major concern in supervised techniques. Overfitting refers to the phenomenon when a network starts learning specific patterns of the input such that it fits well on the training data but leads to poor generalization abilities on unseen data. The accessibility of enormous quantity of data limits the field of medical domain research. This paper focuses on utility of data augmentation (DA) techniques, which is a well-recognized solution to the problem of limited data. The experiments were performed on the Brain Tumor Segmentation (BraTS) dataset which is available online. The results signify that different DA approaches have upgraded the accuracies for segmenting brain tumor boundaries using CNN based model.

This paper presents an automatic diagnosis system for the tumor grade classification through magnetic resonance imaging (MRI). The diagnosis system involves a region of interest (ROI) delineation using intensity and edge magnitude based multilevel thresholding algorithm. Then the intensity and the texture attributes are extracted from the segregated ROI. Subsequently, a combined approach known as Fisher+ Parameter-Free BAT (PFreeBAT) optimization is employed to derive the optimal feature subset. Finally, a novel learning approach dubbed as PFree BAT enhanced fuzzy K-nearest neighbor (FKNN) is proposed by combining FKNN with PFree BAT for the classification of MR images into two categories: High and Low-Grade. In PFree BAT enhanced FKNN, the model parameters, i.e., neighborhood size k and the fuzzy strength parameter m are adaptively specified by the PFree BAT optimization approach. Integrating PFree BAT with FKNN enhances the classification capability of the FKNN. The diagnostic system is rigorously evaluated on four MR images datasets including images from BRATS 2012 database and the Harvard repository using classification performance metrics. The empirical results illustrate that the diagnostic system reached to ceiling level of accuracy on the test MR image dataset via 5-fold cross-validation mechanism. Additionally, the proposed PFree BAT enhanced FKNN is evaluated on the Parkinson dataset (PD) from the UCI repository having the pre-extracted feature space. The proposed PFree BAT enhanced FKNN reached to an average accuracy of 98% and 97.45%. with and without feature selection on PD dataset. Moreover, solely to contrast, the performance of the proposed PFree BAT enhanced FKNN with the existing FKNN variants the experimentations were also done on six other standard datasets from KEEL repository. The results indicate that the proposed learning strategy achieves the best value of accuracy in contrast to the existing FKNN variants.

In telecardiology applications, research has been underway to protect the patient's confidential information from unauthorized access using Electrocardiogram (ECG) steganography or encryption approaches. A novel Fused Coupled Chaotic Map (FCCM) structure based integrated embedding-then-encryption approach is proposed for patient's confidential data embedding in ECG and its subsequent encryption. The proposed ECG embedding-then-encryption algorithm consists of four phases: confidential data ciphering, LSB embedding, substitution, and permutation. Four FCCMs under different initial conditions and control parameters undergoes level shifting or sorting based quantization. The updation of control parameters made to be ECG content aware of security enhancement. The generated {FCCMi,: i ∈ (1–4)} are applied to (i) cipher confidential data, (ii) craft the embedding locations to hide segmented-ciphered confidential data, (iii) substitution and (iv) permutation processes for data embedding-then-encryption of ECG. The performance of the proposed method was evaluated over MIT-BIH Arrhythmia and self-recorded ECG (lead-II) database. The test suites NIST-800 and 0-1 are employed to check randomness of chaotic sequences. The PRD, PRDN, PRD1024, RMS, SNR, PSNR, WWPRD, WEDD, and BERs were 0.04, 1.05, 0.75, 5.55 × 10−4, 41.21, 70.17, 3.02 × 10−2, 5.91 × 10−3, and 0.00 for MIT-BIH Arrhythmia ECG database (Lead-II) at payload 2.4Kb. Performance measures for Self-recorded dataset are 0.04, 2.62, 2.62, 5.49 × 10−4, 32.16, 68.72, 9.36 × 10−2, 1.93 × 10−2, and 0.00 respectively. Experimental results, security analysis and comparison with existing state-of-art methods prove the effectiveness of the proposed approach for assured patient's confidential data security. Furthermore, the efficiency of the proposed approach has also been evaluated on compressed ECG.

This paper presents X-Band wide bandwidth coaxial cavity band pass filter for spacecraft checkout applications. Mostly coaxial cavity filter is used for designing narrow bandwidth filter, less than 1 percent bandwidth but in this paper an attempt has been made to design and develop a compact size filter with more than 1 percent bandwidth. The main advantage of this designed filter is its wide bandwidth and excellent harmonic rejection performance. The realized filter has 1.1 dB insertion loss at center frequency of 8.2 GHz, better than 18 dB return loss over a passband of 7.96 GHz to 8.44 GHz and 60 dB rejection at 720 MHz away from center frequency. Achieved results of proposed filter are in good agreement with the design specifications.

Parkinson’s disease is perhaps the most well-known neurodegenerative disorder that mainly occurs due to the loss of dopamine-producing neurons and consists of motor/non-motor symptoms. The progression of the symptoms is often varying from one person to another to the diversity of the disease. The condition causes a huge burden both on those affected, as well as their families. Accurate diagnosis is critical and challenging but still, no specific diagnostic process is available. The computer-aided diagnosis techniques of signalling and imaging processing are very helpful in the prediction and classification of PD. This review gives a brief description of different methods of classification for early detection and also highlights the most profitable research directions by focusing on continuous monitoring patterns of daily activities, interactions, and routine that may provide the data on status changes, clinical management, and controlling self-correction

Diabetic retinopathy is a globally rising disease and needs to be taken in concern. It is the problem with vision of diabetic patients due to a disease in the retina of diabetic patients.Diabetic patients have high glucose level in the blood.Our major concern is to predict the disease at early stages.The studies focusses on the modern techniques used in image processing digitally.It also puts a stress on patches classification used for the examination and prediction of diabetic retinopathy and judge the accuracy,senstivity of dataset.

This paper aims at presenting a complete picture of advances till now in the field of computer-aided detection of Pulmonary Tuberculosis using Chest X-ray Images. Advances are analyzed in chronological order as they happen and are divided into three phases in which technology shifted into new paradigms. Study concludes that although techniques that use Machine learning based methods for segmentation and classification are prevailing for the moment in terms of flexibility for very particular feature extraction in borderline cases where probabilistic methods can be tweaked according to requirements and accuracy, Deep Convolutional Neural Network based technique will secure higher standings as the computational capability and dataset management improves. Finally, briefly an attempt at using visualization techniques for borderline cases is discussed.

A significant analysis is routine for Brain Tumor patients and it depends on accurate segmentation of Region of Interest. In automatic segmentation, field deep learning algorithms are attaining interest after they have performed very well in various ImageNet competitions. This review focuses on state-of-the-art Deep Learning Algorithms which are applied to Brain Tumor Segmentation. First, we review the methods of brain tumor segmentation, next the different deep learning algorithms and their performance measures like sensitivity, specificity and Dice similarity Coefficient (DSC) are discussed and Finally, we discuss and summarize the current deep learning techniques and identify future scope and trends.

In modern time, emotions affect people in many aspects of life. Long-term emotional problems lead to mental and physical problems such as depression. Wrist pulse contains the information regarding the physiological and pathological state of an individual. Overall mental and physical status of human can be checked with the wrist pulse through pulse examination or nadi parikshan. This paper discusses the parameters of wrist pulse in terms of Vata, Pitta, and Kapha energies for different emotions (anger, calm, fear). Manually, Vata, Pitta, and Kapha are sensed by the medical practitioners through three fingers. Here the pulse data is being acquired through the optical pulse sensor interfaced with the ATMEGA328 microcontroller. Signal processing techniques as approximate entropy, sample entropy, augmentation index, pulse rate are applied in MATLAB to extract the features of Vata, Pitta, and Kapha to recognize the emotions. It is found that Vata rate is high for fear emotion while Pitta rate is high for anger emotion and for calmness emotion, Kapha rate is high. Approximate entropy, sample entropy, augmentation index are more for anger emotion than other emotions.

Wrist pulse signal is important in determining the health status of an individual as it contains the pathological information related to human body. Nadi parikshan or pulse examination is done to obtain the energies (doshas) that are present in every human being. The energies are named as Vata, Pitta and Kapha which are collectively called as bio-elements or Tridosha. Person's physical constitution (Prakruti) is decided by Vata, Pitta and Kapha which occur in unique combination in every individual. The balance between Vata, Pitta and Kapha influence the physiological features of the individual as body weight, body size, eyes, texture and color of hair and skin. This research is intended to find the dominance of the energy based on the body weight of an individual and associated features. In this research, two groups of BMI are considered viz non-obese group ( $\text{BMI} ), obese ( $\text{BMI} > 25$ ). The Data is collected with the help of pulse sensor interfaced with ATMEGA328 microcontroller and features are extracted using the signal processing techniques in MATLAB. It is found that Kapha Rate is high in obese group and small variation in Pitta Rate is also observed for this group. Approximate Entropy and Sample Entropy of the obese group is less than the non-obese group while Augmentation Index of the obese group is more than the non-obese group.

The most important factor involved in heart rate variability (HRV) analysis is cardiac input signal, which is achieved in the form of electrocardiogram (ECG). The ECG signal is used for identifying many electrical defects associated with the heart. In this chapter, many issues involved while ECG recording such as type of the recording instrument, various sources of noise, artifacts, and electrical interference from surroundings is presented. Most importantly, this chapter comprises the details about the experimental protocols followed while ECG recording. Also, the brief overview of medical tourism as well as various interpolation methods used for pre-processing of RR intervals are presented in this chapter.

In this Chapter, a MATLAB-based approach is presented for compression of Electrocardiogram (ECG) data. The methodology employs in three different domains namely direct, transformed and parameter extraction methods. The selected techniques from direct ECG compression methods are TP, AZTEC, Fan, and Cortes. Moreover selected techniques from transformed ECG compression methods are Walsh Transform, DCT, and Wavelet transform. For each of the technique, the basic implementation of the algorithm was explored, and performance measures were calculated. All 48 records of MIT-BIH arrhythmia ECG database were employed for performance evaluation of various implemented techniques. Moreover, based on requirements, any basic techniques can be selected for further innovative processing that may include the lossless encoding.

The present paper proposes a complexity sorting and coupled chaotic map mutation mechanism for compression-then-encryption of the Electrocardiogram (ECG) signals. The compressed-then-encrypted ECG is wirelessly transmitted using orthogonal frequency division multiplexing scheme modified to perform impair sample correction. The compression based on complexity sorting involves following steps: Beat Detection, 2D ECG array formation, Period Normalization, Dc Equalization, Complexity Sorting, Codec Quantization and JPEG2000 codec. The 2D compression results in reduced memory requirements for the clinical data storage. The coupled chaotic based on mutation mechanism for ECG encryption randomizes the ECG array to prevent the attackers from deducing the confidential information from it. The compressed-then-encrypted ECG bitstream is transmitted through the Rayleigh fading wireless channel. At the receiver end, the erroneous sample is corrected by moving median filtering (MMF) mechanism to reduce Percentage Root mean square Difference (PRD). The 2D compressed and the encrypted ECG remains diagnosable after reconstruction. The average compressor metrics Compression Ratio (CR), PRD, and Quality Score (QS) were 72.81 ± 15.90, 2.57 ± 1.68%, and 44.36 ± 29.92 respectively on MIT-BIH Arrhythmia database. Furthermore, 2D compressed ECG is effectually encrypted as validated by the histogram analysis, Information Entropy(En), Entropy Score (ES), and the correlation coefficient analysis. The En, ES, and the correlation values were equal to 7.996, 0.999, and 0.0042 respectively for an 8-bit quantization resolution. Moreover, due to MMF approach at the particular Channel Signal to Noise Ratio (SNRc = 15 dB) & BER = 10−2, the PRD reduces from 45% (without erroneous sample correction) to 2.2% (with erroneous ECG sample correction).

The electrocardiogram (ECG) non-invasively monitors the electrical activities of the heart. During the process of recording and transmission, ECG signals are often corrupted by various types of noises. Minimizations of these noises facilitate accurate detection of various anomalies. In the present paper, Alexander fractional differential window (AFDW) filter is proposed for ECG signal denoising. The designed filter is based on the concept of generalized Alexander polynomial and the R-L differential equation of fractional calculus. This concept is utilized to formulate a window that acts as a forward filter. Thereafter, the backward filter is constructed by reversing the coefficients of the forward filter. The proposed AFDW filter is then obtained by averaging of the forward and backward filter coefficients. The performance of the designed AFDW filter is validated by adding the various type of noise to the original ECG signal obtained from MIT-BIH arrhythmia database. The two non-diagnostic measure, i.e., SNR, MSE, and one diagnostic measure, i.e., wavelet energy based diagnostic distortion (WEDD) have been employed for the quantitative evaluation of the designed filter. Extensive experimentations on all the 48-records of MIT-BIH arrhythmia database resulted in average SNR of 22.014 ± 3.806365, 14.703 ± 3.790275, 13.3183 ± 3.748230; average MSE of 0.001458 ± 0.00028, 0.0078 ± 0.000319, 0.01061 ± 0.000472; and average WEDD value of 0.020169 ± 0.01306, 0.1207 ± 0.061272, 0.1432 ± 0.073588, for ECG signal contaminated by the power line, random, and the white Gaussian noise respectively. A new metric named as morphological power preservation measure (MPPM) is also proposed that account for the power preservance (as indicated by PSD plots) and the QRS morphology. The proposed AFDW filter retained much of the original (clean) signal power without any significant morphological distortion as validated by MPPM measure that were 0.0126, 0.08493, and 0.10336 for the ECG signal corrupted by the different type of noises. The versatility of the proposed AFDW filter is also validated by its application on the ECG signal from MIT-BIH database corrupted by the combination of the noises as well as on the real noisy ECG signals are taken from MIT-BIH ID database. Furthermore, the comparative study has also been done between the proposed AFDW filter and existing state of the art denoising algorithms. The results clearly prove the supremacy of our proposed AFDW filter.

Objective. Cardiovascular diseases generate the highest mortality in the globe population, mainly due to coronary artery disease (CAD) like arrhythmia, myocardial infarction and heart failure. Therefore, an early identification of CAD and diagnosis is essential. For this, we have proposed a new approach to detect the CAD patients using heart rate variability (HRV) signals. This approach is based on subspaces decomposition of HRV signals using multiscale wavelet packet (MSWP) transform and entropy features extracted from decomposed HRV signals. The detection performance was analyzed using Fisher ranking method, generalized discriminant analysis (GDA) and binary classifier as extreme learning machine (ELM). The ranking strategies designate rank to the available features extracted by entropy methods from decomposed heart rate variability (HRV) signals and organize them according to their clinical importance. The GDA diminishes the dimension of ranked features. In addition, it can enhance the classification accuracy by picking the best discerning of ranked features. The main advantage of ELM is that the hidden layer does not require tuning and it also has a fast rate of detection.Methodology. For the detection of CAD patients, the HRV data of healthy normal sinus rhythm (NSR) and CAD patients were obtained from a standard database. Self recorded data as normal sinus rhythm (Self_NSR) of healthy subjects were also used in this work. Initially, the HRV time-series was decomposed to 4 levels using MSWP transform. Sixty two features were extracted from decomposed HRV signals by non-linear methods for HRV analysis, fuzzy entropy (FZE) and Kraskov nearest neighbour entropy (K-NNE). Out of sixty-two features, 31 entropy features were extracted by FZE and 31 entropy features were extracted by K-NNE method. These features were selected since every feature has a different physical premise and in this manner concentrates and uses HRV signals information in an assorted technique. Out of 62 features, top ten features were selected, ranked by a ranking method called as Fisher score. The top ten features were applied to the proposed model, GDA with Gaussian or RBF kernal + ELM having hidden node as sigmoid or multiquadric. The GDA method transforms top ten features to only one feature and ELM has been used for classification.Results. Numerical experimentations were performed on the combination of datasets as NSR-CAD and Self_NSR- CAD subjects. The proposed approach has shown better performance using top ten ranked entropy features. The GDA with RBF kernel + ELM having hidden node as multiquadric method and GDA with Gaussian kernel + ELM having hidden node as sigmoid or multiquadric method achieved an approximate detection accuracy of 100% compared to ELM and linear discriminant analysis (LDA)+ELM for both datasets. The subspaces level-4 and level-3 decomposition of HRV signals by MSWP transform can be used for detection and analysis of CAD patients.

Objective The aims of this study, to investigate the interaction among heart rate variability (HRV), respiratory, systolic arterial blood pressure variability (SABPV) and systolic arterial pressure interval variability (APIV) signals for understanding of cardiovascular control. Methods In this study, three methods referred as adaptive continuous Morlet wavelet transform (ADCMWT), adaptive Stockwell transform (ADST) and adaptive modified Stockwell transform (ADMST) was used to assess the accuracy (AC) of time-varying spectral coherence (TVSC). The adaptation of these estimators was based on maximum energy concentration measurement. The capability to correct temporal localization of time–frequency regions was validated on synthetic time series data modeled as dynamic characteristics of cardiovascular signals. Results The results on synthetic simulated data show that the ADCMWT method allows for the temporal localization of the time–frequency regions with higher accuracy (AC > 96.074% for SNR ≥ 0 dB), compared to ADST (AC > 90.71% for SNR ≥ 0 dB) and ADMST (AC > 84.45% for SNR ≥ 5 dB). Further, the ADCMWT was applied to real cardiovascular data obtained from Fantasia standard data base and grouped as, 8 young subjects (4M + 4F, age range 23–32) and 8 elderly subjects (4M + 4F, age range 70–82) for estimating the TVSC in low frequency (LF) band (0.04 Hz–0.15 Hz) and high frequency (HF) band (0.15 Hz–0.4 Hz) of HRV spectrum. The global result depict that the median value of interquartile range of coherency between HRV-SABPV and HRV-APIV signals in LF and HF band were significantly ( p = 0.00001 ) lower in elderly group subjects compared to young group subjects. The coupling between HRV-Respiratory signals in LF band was not significantly affected with the aging of healthy subjects. However, this coupling in HF band significantly reduced in elderly compare to young group subjects ( p = 0.0125 ). Conclusion The comparative study shows that the time-varying spectra and accurate localization of coupling between two physiological signals can be affected by energy concentration. The ADCMWT at w 0 = 6 , could be an alternative, possibly more suitable and highly accurate method for assessment and detection of time varying spectral and coherence components of cardiovascular time series.

In the present work, a fused metabolite ratio is proposed that integrates the conventional metabolite ratios in a weighted manner to improve the diagnostic accuracy of glioma brain tumor categorization. Each metabolite ratio is weighted by the value generated by the Fisher and the Parameter-Free BAT (PFree BAT) optimization algorithm. Here, feature fusion is formulated as an optimization problem with PFree BAT optimization as its underlying search strategy and Fisher Criterion serving as a fitness function. Experiments were conducted on the magnetic resonance spectroscopy (MRS) data of 50 subjects out of which 27 showed low-grade glioma and rest presented high-grade. The MRS data was analyzed for the peaks. The conventional metabolite ratios, i.e., Choline/N-acetyl aspartate (Cho/NAA), Cho/Creatine (Cho/Cr), were quantitated using peak integration that exhibited difference among the tumor grades. The difference in the conventional metabolite ratios was enhanced by the proposed fused metabolite ratio that was duly validated by metrics of sensitivity, specificity, and the classification accuracy. Typically, the fused metabolite ratio characterized low-grade and high-grade with a sensitivity of 96%, specificity of 91%, and an accuracy of 93.72% when fed to the K-nearest neighbor classifier following a fivefold cross-validation data partitioning scheme. The results are significantly better than that obtained by the conventional metabolites where an accuracy equal to 80%, 87%, and 89% was attained. Prominently, the results using the fused metabolite ratio show a surge of 4.7% in comparison to Cho/Cr + Cho/NAA + NAA/Cr. Moreover, the obtained results are better than the similar works reported in the literature.

In the present paper, a hybrid multilevel thresholding technique that combines intuitionistic fuzzy sets and tsallis entropy has been proposed for the automatic delineation of the tumor from magnetic resonance images having vague boundaries and poor contrast. This novel technique takes into account both the image histogram and the uncertainty information for the computation of multiple thresholds. The benefit of the methodology is that it provides fast and improved segmentation for the complex tumorous images with imprecise gray levels. To further boost the computational speed, the mutation based particle swarm optimization is used that selects the most optimal threshold combination. The accuracy of the proposed segmentation approach has been validated on simulated, real low-grade glioma tumor volumes taken from MICCAI brain tumor segmentation (BRATS) challenge 2012 dataset and the clinical tumor images, so as to corroborate its generality and novelty. The designed technique achieves an average Dice overlap equal to 0.82010, 0.78610 and 0.94170 for three datasets. Further, a comparative analysis has also been made between the eight existing multilevel thresholding implementations so as to show the superiority of the designed technique. In comparison, the results indicate a mean improvement in Dice by an amount equal to 4.00% (p

In this paper, a fractional differential zero phase (FDZP) 2D filter is constructed which is based on the technique of zero-phase filtering utilizing the concept of R–L integral with fractional differentiation. The constructed filter mask is used to denoise an image with the forward-backward processing which gives high robustness for images corrupted with Gaussian noise with varying degree of standard deviation and with speckle noise. The proposed FDZP 2D filter has high edge preserving capability that is also quantitatively evaluated using the peak signal-to-noise ratio (PSNR) performance measure. From the quantitative analysis, the PSNR value is much better than those obtained using Gaussian smoothing, Alexander fractional differential (AFD), Alexander fractional integral (AFI), anisotropic diffusion and kuan filters for standard and ultrasonic image of liver, respectively. The experiments illustrate that the improvements achieved are compatible with other existing filters.

In the present study, a new feature named as density measure is proposed for classification of the glioma brain tumour magnetic resonance (MR) image into low and high-grade categories. The proposed feature is derived using improved complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and Hilbert transformation technique. The proposed feature uses the difference signal created by mapping of the fluid attenuation inversion recovery segmented region onto T1 and T1-contrast-enhanced MR images. This difference signal is decomposed into various intrinsic mode functions (IMFs) using improved CEEMDAN algorithm that effectively captures the texture variations existing in both tumour groups. Then, the Hilbert transformation of resulting IMF is computed that provides the analytic signal representation, thereby giving a better visualisation of the texture difference. The proposed feature is calculated from this analytic signal representation at 97% confidence level. Subsequently, this feature is utilised to calculate a quantitative metric for tumour classification. The proposed metric has been validated on 120 tumorous images taken from brain tumour image segmentation benchmark (BRATS) 2012 dataset and the images taken from Harvard Medical School repository. The results illustrate that proposed metric yields an overall classification accuracy of 100% which is better than existing state-of-art works.

The present paper proposes a novel feature selection technique for the MR brain tumor image classification that aims to choose the optimal feature subset with maximum discriminatory ability in the minimum amount of time. It is based on the fusion of the Fisher and the parameter-free Bat (PFree Bat) optimization algorithm. As the conventional Bat algorithm is bad at exploration, a modification is proposed that guides the Bat by the pulse frequency, global best and the local best position. This improved version of Bat referred to as the PFree Bat algorithm eliminates the velocity equation and directly updates the Bat position. Subsequently, this method in conjunction with the Fisher criteria has been used to select the best set of features for brain tumor classification. The chosen features are then fed to the commonly used least square (LS) support vector machine (SVM) classifier to categorize the area of interest into the high or low grade. For the evaluation of the proposed attribute selection method, tenfold cross-validation has been conducted on a set of 95 ROIs taken from the BRATS 2012 dataset. On an extensive comparison with the other hybrid approaches, the proposed approach brought about the 100% recognition rate in the smallest amount of time. Furthermore, an integrated index is proposed that uniquely identifies the best performing algorithm, taking into account the accuracy, number of features and the computational time. For the fair comparison, the performance of the proposed method has also been examined on breast cancer dataset taken from UCI repository. The obtained results validate that the designed algorithm has better average accuracy than existing state-of-the-art works.

Multilevel thresholding is one of the most popular image segmentation techniques due to its simplicity and accuracy. Most of the thresholding approaches use either the histogram of an image or information from the grey-level co-occurrence matrix (GLCM) to compute the threshold. The medical images like MRI usually have vague boundaries and poor contrast. So, segmenting these images using solely histogram or texture attributes of GLCM proves to be insufficient. This paper proposes a novel multilevel thresholding approach for automatic segmentation of tumour lesions from magnetic resonance images. The proposed technique exploits both intensity and edge magnitude information present in image histogram and GLCM to compute the multiple thresholds. Subsequently, using both attributes, a hybrid fitness function has been formulated which can capture the variations in intensity and the edge magnitude present in different tumour groups effectively. Mutation-based particle swarm optimization (MPSO) technique has been used to optimize the fitness function so as to mitigate the problem of high computational complexity existing in the exhaustive search methods. Moreover, MPSO has better exploration capabilities as compared to conventional particle swarm optimization. The performance of the devised technique has been evaluated and compared with two other intensity- and texture-based approaches using three different measures: Jaccard, Dice and misclassification error. To compute these quantitative metrics, experiments were conducted on a series of images, including low-grade glioma tumour volumes taken from brain tumour image segmentation benchmark 2012 and 2015 data sets and real clinical tumour images. Experimental results show that the proposed approach outperforms the other competing algorithms by achieving an average value equal to 0.752, 0.854, 0.0052; 0.648, 0.762, 0.0177; 0.710, 0.813, 0.0148 and 0.886, 0.937, 0.0037 for four different data sets.

Cardiac arrhythmia, being the key sign regarding heart disease monitored by ECG signals. By carefully analysing ECG signals, we can determine various classes of arrhythmia. PVC is ordinary form of arrhythmia. Diagnosis of PVC ectopic beats is done by using ECG signals which is crucial for the prognosis of probable heart failure. The strategy propounded in this article for PVCs detection is Genetic based SVM. The key features like QRS complex width, Form Factor and RR interval of an ECG signal are extracted and further the parameters of SVM are optimized by using Genetic Algorithm (GA). Experiments with different inputs are supervised to get optimal solution for PVC detection. On testing MIT physionet database, GSVM performs well in PVC detection with accuracy, sensitivity and specitivity of 99.5%, 98% and 100.

Signal processing technology comprehends fundamental theory and implementations for processing data. The processed data is stored in different formats. The mechanism of electrocardiogram (ECG) steganography hides the secret information in the spatial or transformed domain. Patient information is embedded into the ECG signal without sacrificing the significant ECG signal quality. The chapter contributes to ECG steganography by investigating the Bernoulli's chaotic map for 2D ECG image steganography. The methodology adopted is 1) convert ECG signal into the 2D cover image, 2) the cover image is loaded to steganography encoder, and 3) secret key is shared with the steganography decoder. The proposed ECG steganography technique stores 1.5KB data inside ECG signal of 60 seconds at 360 samples/s, with percentage root mean square difference of less than 1%. This advanced 2D ECG steganography finds applications in real-world use which includes telemedicine or telecardiology.

Multilevel thresholding is segmenting the image into several distinct regions. Medical data like magnetic resonance images (MRI) contain important clinical information that is crucial for diagnosis. Hence, automatic segregation of tissue constituents is of key interest to clinician. In the chapter, standard entropies (i.e., Kapur and Tsallis) are explored for thresholding of brain MR images. The optimal thresholds are obtained by the maximization of these entropies using the particle swarm optimization (PSO) and the BAT optimization approach. The techniques are implemented for the segregation of various tissue constituents (i.e., cerebral spinal fluid [CSF], white matter [WM], and gray matter [GM]) from simulated images obtained from the brain web database. The efficacy of the thresholding technique is evaluated by the Dice coefficient (Dice). The results demonstrate that Tsallis' entropy is superior to the Kapur's entropy for the segmentation CSF and WM. Moreover, entropy maximization using BAT algorithm attains a higher Dice in contrast to PSO.

In this chapter, the BAT-optimized fuzzy k-nearest neighbor (FKNN-BAT) algorithm is proposed for discrimination of the electrocardiogram (ECG) beats. The five types of beats (i.e., normal [N], right bundle block branch [RBBB], left bundle block branch [LBBB], atrial premature contraction [APC], and premature ventricular contraction [PVC]) are taken from MIT-BIH arrhythmia database for the experimentation. Thereafter, the features are extracted from five type of beats and fed to the proposed BAT-tuned fuzzy KNN classifier. The proposed classifier achieves the overall accuracy of 99.88%.

In the chapter, dynamic time domain features are extracted in the proposed approach for the accurate classification of electrocardiogram (ECG) heartbeats. The dynamic time-domain information such as RR, pre-RR, post-RR, ratio of pre-post RR, and ratio of post-pre RR intervals to be extracted from the ECG beats in proposed approach for heartbeat classification. These four extracted features are combined and fed to k-nearest neighbor (k-NN) classifier with tenfold cross-validation to classify the six different heartbeats (i.e., normal [N], right bundle branch block [RBBB], left bundle branch block [LBBB], atrial premature beat [APC], paced beat [PB], and premature ventricular contraction[PVC]). The average sensitivity, specificity, positive predictivity along with overall accuracy is obtained as 99.77%, 99.97%, 99.71%, and 99.85%, respectively, for the proposed classification system. The experimental result tells that proposed classification approach has given better performance as compared with other state-of-the-art feature extraction methods for the heartbeat characterization.

The primary objective of this chapter is to analyze the existing tools and techniques for medical data security. Typically, medical data includes either medical signals such as electrocardiogram, electroencephalogram, electromyography, or medical imaging like digital imaging and communications in medicine, joint photographic experts group format. The medical data are sensitive, subject to privacy preservation, and data access rights. Security in e-health field is an integrated concept which includes robust combination of confidentiality, integrity, and availability of medical data. Confidentiality ensures the data is inaccessible to unauthorized access. Integrity restricts the alteration in data by the unauthorized user. Whereas availability provides the readiness of the data when needed by the authorized user. Additionally, confidentiality, integrity and availability, accountability parameter records the back action list which answers the why, when, what, and whom data is accessed. The selected tools and techniques used in medical data security in e-health applications is discussed.

Electrocardiogram (ECG) is one of the best representatives of physiological signal that provides the state of the autonomic nervous system, primarily responsible for the cardiac activity. The ECG data compression plays a significant role in localized digital storage or efficient communication channel utilization in telemedicine applications. The lossless and lossy compression system’s compressor efficiency depends on the methodologies used for compression and the quality measure used to evaluate distortion. Based on domain ECG, data compression can be performed either one-dimensional (1D) or two-dimensional (2D) for utilization of inter and inter with intra beat correlation, respectively. In this paper, a comparative study between 1D and 2D ECG data compression methods was taken out from the existing literature to provide an update in this regard. ECG data compression techniques and algorithms in 1D and 2D domain have their own merits and limitations. Recently, numerous research and techniques in 1D ECG data compression have been developed, including direct and transformed domain. Additionally, 2D ECG data compression research is reported based on period normalization and complexity sorting in recent times. Finally, several practical issues highlight the assessment of reconstructed signal quality and performance comparisons with an average comparative of exhaustive existing 1D and 2D ECG compression methods based on the utilized digital signal processing systems.

Display Omitted Novel level set method for segmenting low contrast cancerous regions is presented.Proposed method uses new stopping function based on edge and phase information.Result indicates the superiority of the proposed method in 2D and 3D CT images. Segmenting low-contrast cancerous regions from Computed Tomography (CT) images is an important task. Region and edge-based active contours fail to perform with such images. Thus, edge-based phase congruent region enhancement is proposed for detecting low-contrast boundaries using new stopping function. The stages of the proposed method are: region separation, region enhancement and Distance Regularised Level Set Evolution (DRLSE) with new stopping function. First, cancerous region is delineated by creating phase map and converting it into an edge map by thresholding. Second, the feature map is created by enhancing features at the boundaries of edge map. The feature map is combined with original image to generate final image. Finally, stopping function is constructed for DRLSE based on the gradient of final image. Experiments were performed on 20 two dimensional and 4 three dimensional CT scans. The proposed method was compared with two active contours. Results prove the superiority of proposed method.

An algorithm is presented for designing a new class of wavelets matched to the Heart Rate Variability (HRV) signals of the menstrual cycle. The proposed wavelets are used to find HRV variations between phases of menstrual cycle. The method finds the signal matching characteristics by minimising the shape feature error using Least Mean Square method. The proposed filter banks are used for the decomposition of the HRV signal. For reconstructing the original signal, the tree structure method is used. In this approach, decomposed sub-bands are selected based upon their energy in each sub-band. Thus, instead of using all sub-bands for reconstruction, sub-bands having high energy content are used for the reconstruction of signal. Thus, a lower number of sub-bands are required for reconstruction of the original signal which shows the effectiveness of newly created filter coefficients. Results show that proposed wavelets are able to differentiate HRV variations between phases of the menstrual cycle accurately than standard wavelets.

Multiscale approximate entropy (MAE) is used to quantify the complexity of a time series as a function of time scale τ. Approximate entropy (ApEn) tolerance threshold selection 'r' is based on either: (1) arbitrary selection in the recommended range (0.1-0.25) times standard deviation of time series (2) or finding maximum ApEn (ApEnmax) i.e., the point where self-matches start to prevail over other matches and choosing the corresponding 'r' (rmax) as threshold (3) or computing rchon by empirically finding the relation between rmax, SD1/SD2 ratio and N using curve fitting, where, SD1 and SD2 are short-term and long-term variability of a time series respectively. None of these methods is gold standard for selection of 'r'. In our previous study [1], an adaptive procedure for selection of 'r' is proposed for approximate entropy (ApEn). In this paper, this is extended to multiple time scales using MAEbin and multiscale cross-MAEbin (XMAEbin). We applied this to simulations i.e. 50 realizations (n = 50) of random number series, fractional Brownian motion (fBm) and MIX (P) [1] series of data length of N = 300 and short term recordings of HRV and SBPV performed under postural stress from supine to standing. MAEbin and XMAEbin analysis was performed on laboratory recorded data of 50 healthy young subjects experiencing postural stress from supine to upright. The study showed that (i) ApEnbin of HRV is more than SBPV in supine position but is lower than SBPV in upright position (ii) ApEnbin of HRV decreases from supine i.e. 1.7324 ± 0.112 (mean ± SD) to upright 1.4916 ± 0.108 due to vagal inhibition (iii) ApEnbin of SBPV increases from supine i.e. 1.5535 ± 0.098 to upright i.e. 1.6241 ± 0.101 due sympathetic activation (iv) individual and cross complexities of RRi and systolic blood pressure (SBP) series depend on time scale under consideration (v) XMAEbin calculated using ApEnmax is correlated with cross-MAE calculated using ApEn (0.1-0.26) in steps of 0.02 at each time scale in supine and upright position and is concluded that ApEn0.26 has highest correlation at most scales (vi) choice of 'r' is critical in interpreting interactions between RRi and SBP and in ascertaining true complexity of the individual RRi and SBP series.

The aim of this study is to propose a new baroreflex sensitivity (BRS) index using improved Hilbert–Huang transform (HHT) using weighted coherence (CW) criterion and apply it to assess baroreflex in supine and standing postures. Improved HHT is obtained by addressing the mode mixing and end effect problems associated with empirical mode decomposition which is a required step in the computation of HHT and thus mitigating the unwanted low frequency component from the power spectrum. This study was first performed on synthetic signals generated using integral pulse frequency model and further extended to real RR interval and systolic blood pressure records of 50 healthy subjects, 20 post acute myocardial infarction patients undergoing postural stress from supine to standing position. Evaluation is also performed on standard EuroBaVar database, comprising of 21 subjects, under supine and standing positions. The results are (i) enhanced values of supine-to-standing low frequency BRS index (α-LF) equal to 1.78 and high frequency BRS index (α-HF) equal to 2.48 are obtained using improved HHT compared to standard HHT (α-LF = 1.54, α-HF = 2.36) and traditional power spectral density (α-LF = 1.55, α-HF = 2.34) for healthy subjects, (ii) there is an increased rate of change of LF/HF power ratios from supine to standing positions, and (iii) number of BRS responses obtained using CW criterion are greater than those obtained by using mean coherence criterion. In conclusion, the new BRS index takes into consideration the non-linear nature of interactions between heart rate variability and systolic blood pressure variability.

In this paper, a computer aided diagnostic (CAD) system for classification of abdomen diseases from computed tomography (CT) images is presented. The methodology used in this paper is to select the most appropriate machine learning technique of segmentation, feature extraction and classification for each module of proposed CAD. The methodology of selecting appropriate machine learning technique for each module of CAD results in accurate and efficient system. Regions of interest are segmented from CT images of tumor, cyst, calculi and normal liver using active contour models, region growing and thresholding. The CAD presented in this research work exploits the discriminating power of features for classifying abdominal diseases. Therefore, feature extraction module extracts statistical texture descriptors using three kinds of feature extraction methods i.e. Gray-Level co-occurrence matrices (GLCM), Discrete Wavelet Transform (DWT) and Discrete Curvelet Transform (DCT). At the next stage, effective and optimum features of ROIs are selected using Genetic Algorithm (GA). Further, Support Vector Machine (SVM) and Artificial Neural Network (ANN) are used to assess the capability of features for classification of diseases of abdomen. The study is performed on 120 CT images of abdomen (30 normal, 30 tumor, 30 cyst and 30 calculi). It is observed from the results that proposed CAD consists of edge based active contour model combined with optimized statistical texture descriptors using DCT along with ANN as classifier achieves the best diagnostic performance of 95.1%. It is also shown in results that proposed CAD achieves highest sensitivity, specificity of 95% and 98% respectively.

This paper presents a quality controlled reconstruction of ECG signal by the formulation of 2D Discrete Cosine Transform (DCT) coefficient and iterative JPEG2000 encoding scheme for Electrocardiogram (ECG) data compression. It employs variable blockwise (L = 32 to 2048 samples) DCT to exploit inter and intra-beat correlation inherent in ECG. The Regula-Falsi method is applied to obtain the optimal threshold for the DCT coefficients, which are stacked row-wise in order to form a 2D DCT coefficient array for the correlation analysis. Afterwards, window-based filtering is applied to retain the significant DCT coefficients with varying window size (ws). Differential pulse code modulation based quantization scheme is then applied to 2D DCT coefficients array in order to make it conformal to JPEG2000 encoding. Finally, iterative JPEG2000 encoding is utilized for achieving the overall predefined reconstruction error. The experimentations were performed on all the 48 records of Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database. The proposed method provides compression ratio = 39.48 & 40.18; percentage root mean square difference = 2.02 & 2.06 and the quality score = 19.45 & 19.56 at ws = 50% & 20% respectively for L = 256. The proposed technique results were also validated on all records of four standard ECG databases including i) MIT-BIH Normal sinus rhythm ii) CU ventricular tachyarrhythmia iii) CSE and iv) PTB Database. Further, a comparative analysis between the proposed method and some recent existing works was also performed. The results clearly demonstrate the superiority of proposed method.

The image steganography is the technique which can provide security to sensitive data. The various methods are designed so far to generate efficient stegno image. The genetic algorithm is applied in the existing system to detect edges from the images and on the detected edges text is inserted to generate final stegno image. In this research work, the stegno image is generated using the GLCM algorithm for the features extraction. The PSO algorithm is applied to detect portion from the image in which we need to insert text. The PCA Algorithm is applied to extract useful features of image. The performance of proposed modal is tested in MATLAB and it is analyzed that proposed technique performs well in terms of PSNR and MSE

The variations in heart rate in various menstrual cycle phases have posed challenge to the cardiac automatic control. The Heart rate variability (HRV) is important parameter to decode the state of the heart during the sensitive period of menstrual cycle. There are limited studies which concentrated on detecting the HRV between menstrual cycle phases. Detecting HRV between phases is important to judge the condition of heart between phases. Thus, the objective is to investigate the effect of change of various phases of the menstrual cycle on autonomic modulations of heart rate using HRV analysis methods. In this paper, linear methods are applied on data set of 105 women recorded for different phases of menstrual cycle. The results concluded that pNN50 and RMSSD are more accurate methods in identifying the variations in hear rate between menstrual vs follicular phase and luteal vs menstrual phase of the menstrual cycle. the pNN50 method is more accurate in detecting the heart rate variations in luteal vs follicular phase.