Biomedical Engineering: Applications, Basis and Communications

Many of the diseases that are related to the nervous system are serious and can lead to pain, headaches, and many other symptoms. Magnetic resonance imaging (MRI), which produces detailed medical images for the diagnosis of nervous system-related diseases, offers an advanced approach to the study of these diseases. In this paper, a method for the automatic segmentation of the skull, scalp, and brain of 3D MRI imagery is proposed to help doctors diagnose numerous injuries and diseases. For the proposed method, the 3D MRI dataset was first transformed to 2D axial slices, which were grouped based on the skull shape. Second, the histogram method was used to compute the fitting rectangle boundary around the object. Third, the segmentation of the skull was performed using the proposed adaptive region growing method, depending on the slice groups. Finally, the brain and the scalp were segmented. The proposed method was tested and validated on 20 subjects from the BrainWeb database and seven adults from the Neurodevelopmental MRI database with the use of the tissues for the ground truth data. The performance of the proposed method was evaluated using the Dice percentage and a comparison with the other methods.

The diagnosis of benign and malignant breast cancer is a challenging issue today. Breast cancer is the most common cancer that women suffer from. The sooner the cancer is detected, the easier and more successful it is to treat it. The most common diagnostic method is the mammography of a simple radiographic picture of the chest. The use of image processing techniques and identifying patterns in the detection of breast cancer from mammographic images reduces human errors in detecting tumors, and speeds up diagnosis time. Artificial Neural Network (ANN) has been widely used to detect breast cancer, and has significantly reduced the percentage of errors. Therefore, in this paper, Convolution as Neural Network (CNN), which is the most effective method, is used for the detection of various types of cancers. This study presents a Multiscale Convolutional Neural Network (MCNN) approach for the classification of tumors. Based on the structure of MCNN, which presents mammography picture to several deep CNN with different size and resolutions, the classical handcrafted features extraction step is avoided. The proposed approach gives better classification rates than the classical state-of-the-art methods allowing a safer Computer-Aided Diagnosis of pleural cancer. This study reaches the diagnosis accuracy of $97\pm 0.3$ using multiscale convolution technique which reveals the efficient proposed method.

In this paper, the Oppositional Whale Optimization Algorithm (OWOA) is applied to Adaptive Noise Canceller (ANC) for the filtering of Electroencephalography/Event-Related Potentials (EEG/ERP) signals. Performance of ANC will be improved by calculating the optimal weight value and proposed OWOA technique is used to update weight value. Adaptive filter’s noise reduction capability has been tested through consideration of White Gaussian Noise (WGN) over contaminated EEG signals at various SNR levels ($-$10dB, $-$15dB and $-$20dB). The performance of the proposed OWOA algorithm is assessed in terms of Signal to Noise Ratio (SNR) in dB, mean value, and the correlation between resultant and input ERP. In this work, ANCs are also implemented by utilizing conventional gradient-based techniques like Recursive Least Square (RLS), Least Mean Square (LMS) and other optimization algorithms such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and WOA techniques. In average cases of noisy environment, comparative analysis shows that the proposed OWOA technique provides higher SNR value and significantly lower mean, and correlation as compared to gradient-based and swarm-based techniques. The comparative results show that extracting the desired EEG component is more effective in the proposed OWOA method. So, it has seen that OWOA-based noise reduction technique removing the artifacts and improving the quality of EEG signals significantly for biomedical analysis.

The oxidised hyaluronic acid/adipic acid dihydrazide hydrogel (oxi-HA/ADH) is a biocompatible and biodegradable hydrogel that has been applied in tissue engineering in musculoskeletal medicine. The present study investigated the feasibility of using oxi-HA/ADH as a drug carrier to incorporate cytoprotective medications for possible tendinopathy treatment. The cytocompatibility of oxi-HA/ADH, epigallocatechin gallate (EGCG) and piracetam were tested. Medications were incorporated in oxi-HA/ADH in the cross-linking process and the drug release profile was tested in vitro. Sprague-Dawley rats Achilles tendon-derived cells (TDCs) were isolated and cultured in a self-designed bioreactor exerting 8% cyclic strain at 0.5Hz for 8h. TDCs were treated with EGCG, piracetam, or no medication during mechanical loading. Gene expression of type I and type III collagen, Pparg, Sox9, and Runx2 (representing adipogenic, chondrogenic, and osteogenic lineages respectively) were analyzed by quantitative polymerase chain reaction. EGCG and piracetam showed no cytotoxicity at therapeutic range. The cumulative release profile of the drug-loaded oxi-HA/ADH showed a two-phase release of medication, a burst of 50–60% during the first 24h, followed by a steady release over 1–7 days. The decreased type I collagen expression and increased non-tenocyte lineage genes expression of TDCs upon 8% cyclic strain is reversed in the EGCG and piracetam treated groups, compared with non-treated group. Oxi-HA/ADH serves as a potential drug carrier that shows sustained-release characteristics. Anti-oxidative and cytoprotective medications, EGCG and piracetam, were able to suppress the aberrant non-tenocyte lineage genes expression of the TDCs upon excessive mechanical loading.

Electromyogram signals have been used for various applications in the healthcare sector for developing various methodologies and techniques in rehabilitation and prosthetics. This paper focuses on the use of EMG signals of trans-radial amputees for developing a myoelectric lower limb prosthesis capable of individual finger movement. The aim of this work is to develop proper hardware and software systems for real-time EMG classification. An improved double thresholding method for onset and offset detections has been developed to ensure its applicability in real-time. The proposed algorithm has been tested with real-time patient EMG signals using a three-lead electrode system from flexor digitalis region of the hand. Around 3000 samples of usable data corresponding to the flexion of each finger (Thumb-553, Index-655, Middle-723, Ring-720, Little-655) were acquired from 10 healthy subjects. The resultant extracted features were classified using various classifiers (KNN, KNN with PCA and LDA) and a comparison was done between the accuracies acquired from a commonly shared dataset against a subject-specific dataset. A robust onset signal processing algorithm enabled the real-time classification of EMG in noisy environments.

This piece of work investigates the application of histogram equalization method to clinical images for noise removal and efficient image enhancement without any information loss. Computed tomographic (CT) images of the abdomen bearing liver tumour are kept under study. Liver exhibits heterogeneous combination of intensities which makes it a challenging task to enhance the liver tumour embedded in the image. Distortion occurs due to the presence of quantum noise in the CT scans and important information of the image is suppressed. The methodology adopted in this paper comprises of two stages. Initially pixel based intensity transformation is adopted for de-noising the background of the image by the selection of appropriate threshold levels. The resultant image gives a noise free background and the foreground features are enhanced. In the next stage histogram equalization filters are applied to the transformed image. The equalization method which gives uniform image enhancement with lesser mean square error (MSE) and increased peak signal to noise ratio (PSNR) is supposed to be an effective method for efficient enhancement of the images. This study deals with the application of histogram equalization methods to CT images which can aid the radiologists for better visualization and diagnosis of the disease.

Detection and diagnosis of glaucoma disease of eye fundus images at early stage is very important as this disorder leads to complete loss of vision if ignored. Usually, 80–90% of glaucoma cases are analyzed manually by ophthalmologists. As the manual analysis varies from one expert to other, diagnosis cannot be effective. Hence, there is a need for automatic assessment of glaucoma disease using computer aided diagnosis (CAD). Many researchers have devised several CAD techniques for glaucoma analysis using various classification techniques. However, most of the classifiers are efficient only for two level classification to detect whether disease is glaucoma or not. But, glaucoma disease has several stages and demands multilevel approaches with high degree of classification accuracy. Among several multiclass methods, literature suggests multiclass support vector technique (MSVM) as a better performing statistical classifier. However, many MSVMS suffer from data loss during training phase. To address this issue, a robust hybrid classification approach consisting of Naïve Bayes binary classifier in the first stage and simplified multiclass support vector machine (Sim-MSVM) in the second stage is proposed in this paper.

Alzheimer’s disease (AD) is an irreversible, progressive disorder that assaults the nerve cells of the brain. It is the most widely recognized kind of dementia among older adults. Apolipoprotein E (APOE), is one of the most common genetic risk factors for AD whose significant association with AD is observed in various genome-wide association studies (GWAS). Single nucleotide polymorphisms (SNPs) are the most common type of genetic variation among individuals. SNPs related to many common diseases like AD. SNPs are recognized as significant biomarkers for this disease, they help in understanding and detecting the disease in its early stages. Detecting SNPs biomarkers associated to the disease with high classification accuracy leads to early prediction and diagnosis. Machine learning techniques are utilized to discover new biomarkers of the disease. Sequential minimal optimization (SMO) algorithm with different kernels, Naive Bayes (NB), tree augmented Naive Bayes (TAN) and K2 learning algorithm have been applied on all genetic data of Alzheimer’s disease neuroimaging initiative phase 1 (ADNI-1)/Whole genome sequencing (WGS) datasets. The highest classification accuracy was achieved using 500 SNPs based on the $p$-value threshold ($p$-value $<0.05$). In whole genome approach ADNI-1, results revealed that NB and K2 learning algorithms scored an overall accuracy of 98% and 98.40%, respectively. In whole genome approach WGS, NB and K2 learning algorithms scored an overall accuracy of 99.63% and 99.75%, respectively.