Showing papers in "Journal of Medical Imaging and Health Informatics in 2020"

Modelling, Simulation and Optimization of Diagnosis Cardiovascular Disease Using Computational Intelligence Approaches

Abstract: Background: To provide ease to diagnose that serious sickness multi-technique model is proposed. Data Analytics and Machine intelligence are involved in the detection of various diseases for human health care. The computer is used as a tool by experts in the medical field, and the computer-based mechanism is used to diagnose different diseases in patients with high Precision. Due to revolutionary measures employed in Artificial Neural Networks (ANNs) within the research domain in the medical area, which appear to be in the data-driven applications usually described in the domain of health care. Cardio sickness according to name is a type of an ailment that is directly connected to the human heart and blood circulation setup, so it should be diagnosed on time because the delay of diagnosing of that disease may lead the sufferer to death. The research is mainly aimed to design a system that will be able to detect cardiovascular sickness in the sufferer using machine learning approaches. Objective: The main objective of the research is to gather information of the six parameters that is age, chest pain, electrocardiogram, systolic blood pressure, fasting blood sugar and serum cholesterol are used by Mamdani fuzzy expert to detect cardiovascular sickness. To propose a type of device which will be successfully used in overcoming the cardiovascular diseases. This proposed model Diagnosis Cardiovascular Disease using Mamdani Fuzzy Inference System (DCD-MFIS) shows 87.05 percent Precision. To delineate an effective Neural Network Model to predict with greater precision, whether a person is suffering from cardiovascular disease or not. As the ANN is composed of various algorithms, some will be handed down for the training of the network. The main target of the research is to make the use of three techniques, which include fuzzy logic, neural network, and deep machine learning. The research will employ the three techniques along with the previous comparisons, and given that, the results will be compared respectively. Methods: Artificial neural network and deep machine learning techniques are applied to detect cardiovascular sickness. Both techniques are applied using 13 parameters age, gender, chest pain, systolic blood pressure, serum cholesterol, fasting blood sugar, electrocardiogram, exercise including angina, heart rate, old peak, number of vessels, affected person and slope. In this research, the ANN-based research is one of the algorithms collections, which is the detection of cardiovascular diseases, is proposed. ANN constitutes of many algorithms, some of the algorithms are employed in the paper for the training of the network used, to achieve the prediction ratio and in contrast of the comparison of the mutual results shown. Results: To make better analysis and consideration of the three frameworks, which include fuzzy logic, ANN, Deep Extreme Machine Learning. The proposed automated model Diagnosis Cardiovascular Disease includes Fuzzy logic using Mamdani Fuzzy Inference System (DCD-MFIS), Artificial Neural Network (DCD–ANN) and Deep Extreme Machine Learning (DCD–DEML) approach using back propagation system. These frameworks help in attaining greater precision and accuracy. Proposed DCD Deep Extreme Machine Learning attains more accuracy with previously proposed solutions that are 92.45%. Conclusion: From the previous comparisons, the propose automated Diagnosis of Cardiovascular Disease using Fuzzy logic, Artificial Neural Network, and deep extreme machine learning approaches. The automated systems DCDMFIS, DCD–ANN and DCD–DEML, the framework proposed as effective and efficient with 87.05%, 89.4% and 92.45 % success ratios respectively. To verify the performance which lies in the ANNs and computational analysis, many indicators determining the precise performance were calculated. The training of the neural networks is made true using the 10 to 20 neurons layers which denote the hidden layer. DEML reveals and indicates a hidden layer containing 10 neurons, which shows the best result. In the last, we can conclude that after making a consideration among the three techniques fuzzy logic, Artificial Neural Network and Proposed DCD Deep Extreme Machine, the Proposed DCD Deep Extreme Machine Learning based solution give more accuracy with previously proposed solutions that are 92.45%.

Classification of Alzheimer’s Disease via Eight-Layer Convolutional Neural Network with Batch Normalization and Dropout Techniques

Abstract: More than 35 million patients are suffering from Alzheimer’s disease and this number is growing, which puts a heavy burden on countries around the world. Early detection is of benefit, in which the deep learning can aid AD identification effectively and gain ideal results. A novel eight-layer convolutional neural network with batch normalization and dropout techniques for classification of Alzheimer’s disease was proposed. After data augmentation, the training dataset contained 7399 AD patient and 7399 HC subjects. Our eight-layer CNN-BN-DO-DA method yielded a sensitivity of 97.77%, a specificity of 97.76%, a precision of 97.79%, an accuracy of 97.76%, a F1 of 97.76%, and a MCC of 95.56% on the test set, which achieved the best performance in seven state-of-the-art approaches. The results strongly demonstrate that this method can effectively assist the clinical diagnosis of Alzheimer’s disease.

An Open-Access Long-Term Wearable ECG Database for Premature Ventricular Contractions and Supraventricular Premature Beat Detection

Abstract: Wearable electrocardiogram (ECG) devices can provide real-time, long-term, non-invasive and comfortable ECG monitoring for premature beats (PB) assessment (typically presenting as premature ventricular contractions (PVC) and supraventricular premature beat (SPB)), which may foreshadow stroke or sudden cardiac death. However, the poor quality, introduced by the dry electrode in wearable ECG monitoring system, leads to the inefficient recognition of the existing PB detection technologies. Although many methods can achieve high recognition rate on current widely-used open-access clinical ECG databases, they still fail to work properly on dynamic ECG signals. This study presents an open-access ECG database comprises of 24-hour wearable ECG recordings. The database is used for the 3rd China Physiological Signal Challenge (CPSC 2020), where participants are expected to recognize PVC and SPB from these recordings. All the approved algorithms are evaluated by scoring standards and regulations defined in terms of PVC detection and SPB detection, respectively.

Medical Sports Rehabilitation Deep Learning System of Sports Injury Based on MRI Image Analysis

Abstract: Medical sports rehabilitation deep learning system of sports injury based on MRI image analysis is proposed in this paper. Preparation activities are various body exercises that are purposely performed before physical education, training, and competition. It is a transitional phase from the static state to the moving state of the human body. Preparatory activities can improve the excitability of the central nervous system, improve the ability of the cerebral cortex to analyze and judge movements, and thus make the movement more coordinated and accurate. At the same time prepare activity can also improve the respiratory and circulatory system functions and reduce the muscles, ligaments of the sticky nature and the contraction of muscles for speed and strength, in order to maximize the capacity of the physical movement and injury prevention campaign ready. Therefore, how to use the MRI image to numerically analyze the mentioned task is essential. We integrate the deep learning model to propose the novel image enhancement and recognition model to undertake the task of medical sports rehabilitation system. The experimental result proves the performance is robust.

An Efficient Coding Network Based Feature Extraction with Support Vector Machine Based Classification Model for CT Lung Images

Abstract: Lung cancer is a serious illness affects people all over the globe. To increase the survival rate of patients affected by lung cancer, in advance recognition of lung cancer with effective treatments is important. This study introduces a new deep learning (DL) based feature extraction and classification technique for CT lung images. A DL model using Coding Network (CN) is presented for the extraction of high-level features and classical features. Initially, the convolution neural network is trained as a coding network and the actual pixels are coded into feature vectors for representing the high-level concepts for classification. Next, an extraction of chosen classical features takes place depending upon background knowledge of lung CT images. In addition, an automatic feature fusion takes place to avoid annoying parameter choice. Besides, support vector machine (SVM) model is employed for classify CT lung images in an effective way. For experimentation, a benchmark dataset is utilized to appraise the outcome of the presented CN-SVM model and is validated under several dimensions.

Medical Imaging Fusion Techniques: A Survey Benchmark Analysis, Open Challenges and Recommendations

Abstract: Background—Recent improvements in image processing along with active collaborations of health experts have started an era of inventions in medical imaging. From the last two decades, computer vision empowers computers to analyze the data in bulk using machine learning methods in developing intelligent models. Several techniques are already available in the literature that is capable of learning intricate patterns to generate meaningful output. A set of areas, which researchers have preferably covered are related to contrast stretching, segmentation, feature extraction/fusion, and classification. In the medical domain, feature fusion is an active area, which plays a vital role in the final classification step. Objective—In this review article, our primary objective is to discuss and validate the advantages of feature fusion and present its roles in this domain. The fusion of two images or multiple features gives better results in the form of either detection or classification of infected areas. For this purpose, we discuss a set of techniques used by several researchers in the medical domain. Method—A detailed and comprehensive review of the fusion techniques are presenting. The key challenges and shortcomings of existing image and feature fusion methods are presenting along with the possible future directions. Conclusion—At the end of this review, we conclude that fusion techniques improves the image quality, as well as for salient (infected) regions detection. Moreover, the fusion of medical images increases the overall segmentation accuracy, and feature fusion shows its effects in the final stage of classification.

Applying Deep Learning Technique for Depression Classification in Social Media Text

Abstract: In social media, depression identification could be regarded as a complex task because of the complicated nature associated with mental disorders. In recent times, there has been an evolution in this research area with growing popularity of social media platforms as these have become a fundamental part of people's day-to-day life. Social media platforms and their users share a close relationship due to which the users' personal life is reflected in these platforms on several levels. Apart from the associated complexity in recognising mental illnesses via social media platforms, implementing supervised machine learning approaches like deep neural networks is yet to be adopted in a large scale because of the inherent difficulties associated with procuring sufficient quantities of annotated training data. Because of such reasons, we have made effort to identify deep learning model that is most effective from amongst selected architectures with previous successful record in supervised learning methods. The selected model is employed to recognise online users that display depression; since there is limited unstructured text data that could be extracted from Twitter.

Contrast Enhancement of Low-Contrast Medical Images Using Modified Contrast Limited Adaptive Histogram Equalization

Abstract: The low contrast medical images seriously affect the clinical diagnosis process. To improve the image quality, we propose an effective medical images contrast enhancement technique in this paper. Shear wavelet transformation is used for decomposition of image components into low-frequency and high-frequency. The low-frequency part contrast is adjusted by applying modified contrast limited adaptive histogram equalization (CLAHE). The resultant image is further processed through technique of fuzzy contrast enhancement to maintain the spectral information of an image. Results of the experimentation show that our proposed technique enhance the image contrast up to a good degree while preserving the image details.

Automatic Detection of Diabetic Retinopathy with Statistical Method and Bayesian Classifier

Abstract: Currently, the underlying medical conditions in China lag behind those in urban areas. There are some problems such as lack of resources of primary ophthalmologists and insufficient fundus image of diabetic retinopathy (DR) with markers. To solve the above questions, an automated detection model of diabetic retinopathy based on the statistical method and Naïve Bayesian (NB) classifier is proposed in this paper. Firstly, three sets of texture features are extracted, which are gray-level co-occurrence matrix texture features, different statistical texture features, and gray-level run-length matrix texture features. Secondly, the extracted texture features are used as input of the Naïve Bayesian classifier to classify the fundus images of diabetic retinopathy into three categories. The proposed automatic detection model for diabetic retinopathy is validated by a data set consisting of 568 images from China diabetic retinopathy screening project. The positive predictive accuracy of the system is 93.44%, the sensitivity and specificity are 91.94% and 88.24%, respectively.

Discovering the Predictive Value of Clinical Notes: Machine Learning Analysis with Text Representation

Abstract: Hospital readmission shortly after discharge is threatening to plague the quality of inpatient care. Readmission is a severe episode that leads to increased medical care costs. Federal regulations and early readmission penalties have created an incentive for healthcare facilities to reduce their readmission rates by predicting patients at a high risk of readmission. Scientists have developed prediction models by using rule-based assessment scores and traditional statistical methods, and most have focused on structured patient records. Recently, a few researchers utilized unstructured clinical notes. However, they achieved moderate prediction accuracy by making predictions of a single diagnosis subpopulation via extensive feature engineering. This study proposes the use of machine learning to learn deep representation of patient notes for the identification of high-risk readmission in a hospital-wide population. We describe and train several predictive models (standard machine learning and neural network), to which several setups have not been applied. Results show that complex deep learning models significantly outperform (P < 0.001) conventionally applied simple models in terms of discrimination ability. We also demonstrate a simple feature evaluation using a standard model, which allows the determination of potential clinical conditions/procedures for targeting. Unlike modeling using structured patient information with considerable variability in structure when different templates or databases are adopted, this study shows that the machine learning approach can be applied to prognosticate readmission with clinical free text in various healthcare settings. Using minimum feature engineering, the trained models perform comparably well or better than other predictive models established in previous literature.

Hybrid Kernel Support Vector Machine Classifier and Grey Wolf Optimization Algorithm Based Intelligent Classification Algorithm for Chronic Kidney Disease

Abstract: In the present day, distributed algorithms become more popular due to their diversity in several applications. The prediction and reorganization of medical data required more practice and information. We propose a novel approach feature selection based on efficient chronic kidney disease (CKD) prediction and classification. Primarily, the pre-processing pace will be implemented over the input data. Then, the grey wolf optimization (GWO) algorithm gets executed to choose the optimal features from the pre-processed data. Next, the projected technique exploits the Hybrid Kernel Support Vector Machine (HKSVM) as a classification model to identify the presence of CKD or not. The simulation takes place in MATLAB. The validation of the presented model takes place using a benchmark CKD dataset as of machine learning repository such as UCI under the presence of several measures. New outcome specified that the planned categorization arrangement has surpassed by containing enhanced 97.26% accuracy for kidney chronic dataset when contrasted with existing SVM technique only accomplished 94.77% and fuzzy min–max GSO neural network (FMMGNN) classifier accomplished 93.78%.

Type 2 Machine Learning: An Effective Hybrid Prediction Model for Early Type 2 Diabetes Detection

Abstract: Importance: Diabetes is a chronic disease that can cause long term damage to various parts of the body. To prevent diabetic complications, different attempts integrating machine learning with medicine have been made for building models to predict whether a patient has diabetes or not, but predicting this disease still has room for improvement. Hybrid prediction model presents a novel method and mostly achieve a much better optimal outcome than single classical machine learning algorithms. Objective: To develop a high accuracy model for different onsets of type 2 diabetes prediction. In this way, the integration between clustering and classification techniques can be improved to help detecting diabetes at an earlier stage without deleting observations with missing values and also decrease insignificant features to get the most related features during data collection. Methods: We implement a noise reduction based technique using Kmeans clustering followed by running the Random forest and XGBoost classifiers to extract the unknown hidden features of the dataset and for more accurate results. Results: Prediction accuracy can be observed by benchmarking our model against up-to-date predictive models and common classification algorithms. With an accuracy of 97.53% by 10 fold cross validation, our T2ML model reaches a better accuracy compared with other experiments reported by other researchers in the literature and over various conventional classification algorithms.

An Effective Algorithm for Beat-to-Beat Heart Rate Monitoring from Ballistocardiograms

Abstract: Ballistocardiograms (BCG) is an essential signal for vital sign monitoring. Obtaining the beat-to-beat intervals from BCG signal is of great significance for home-care applications, such as sleep staging, heart disease alerting, etc. The current approaches of detecting beat-to-beat intervals from BCG signals are complex. In this paper, we develop a non-invasive BCG monitoring system, and propose an effective and accurate algorithm for beat-to-beat detection. Firstly, a heartbeat shape is adaptively modeled based on a two-step procedure by taking advantage of the J-peak and the K-valley of BCG signals. Then, forward and backward detections with the criteria of both the morphological distance and the cross-correlation are jointly employed to find the position of each BCG signal, and in turn, to determine the beat-to-beat intervals of BCG signals. The proposed method was validated in at least 90 minutes recording from 10 subjects in various setups. The mean absolute beat-to-beat intervals error was 10.72 ms and on an average 97.93% of the beat-to-beat intervals were detected.

Recognition of Large-Scale ncRNA Data Using a Novel Multitask Cross-Learning 0-Order TSK Fuzzy Classifier

Abstract: Recognizing noncoding ribonucleic acid (ncRNA) data is helpful in realizing the regulation of tumor formation and certain aspects of life mechanisms, such as growth, differentiation, development, and immunity. However, the scale of ncRNA data is usually very large. Using machine learning (ML) methods to automatically analyze these data can obtain more precise results than manually analyzing these data, but the traditional ML algorithms can process only small-scale training data. To solve this problem, a novel multitask cross-learning 0-order Takagi–Sugeno–Kang fuzzy classifier (MT-CL-0-TSK-FC) is proposed that uses a multitask cross-learning mechanism to solve the large-scale learning problem of ncRNA data. In addition, the proposed MT-CL-0-TSK-FC method naturally inherits the interpretability of traditional fuzzy systems and eventually generates an interpretable rulesbased database to recognize the ncRNA data. The experimental results indicate that the proposed MT-CL-0TSK-FC method has a faster running time and better classification accuracy than traditional ML methods.

Dual-Ray Net: Automatic Diagnosis of Thoracic Diseases Using Frontal and Lateral Chest X-rays

Abstract: Computer-aided diagnosis (CAD) is an important work which can improve the working efficiency of physicians. With the availability of large-scale data sets, several methods have been proposed to classify pathology on chest X-ray images. However, most methods report performance based on a frontal chest radiograph, ignoring the effect of the lateral chest radiography on the diagnosis. This paper puts forward a kind of model, Dual-Ray Net, of a deep convolutional neural network which can deal with the front and lateral chest radiography at the same time by referring the method of using lateral chest radiography to assist diagnose during the diagnosis used by radiologists. Firstly, we evaluated the performance of parameter migration to small data after pre-training for large datasets. The data sets for pre-training are chest X-ray 14 and ImageNet respectively. The results showed that pre-training with chest X-ray 14 performed better than with the generic dataset ImageNet. Secondly, We evaluated the performance of the Frontal and lateral chest radiographs in different modes of input model for the diagnosis of assisted chest disease. Finally, by comparing different feature fusion methods of addition and concatenation, we found that the fusion effect of concatenation is better, which average AUC reached 0.778. The comparison results show that whether it is a public or a non-public dataset, our Dual-Ray Net (concatenation) architecture shows improved performance in recognizing findings in CXR images when compared to applying separate baseline frontal and lateral classes.

A Novel Surface Registration for Image-Guided Neurosurgery: Effects of Intervals of Points in Patient Space on Registration Accuracy

Abstract: Surface registration is an important factor in surgical navigation in determining the success rate and stability of surgery by providing the operator with the exact location of a lesion. The problem of surface registration is that point cloud in the patient space is acquired at irregular intervals due to the operator’s tracking speed and method. The purpose of this study is to analyze the effect of irregular intervals of point cloud caused by tracking speed and method on the registration accuracy. For this study, we created the head phantom to obtain a point cloud in the patient space with a similar object to that of a patient and acquired a point cloud in a total of ten times. In order to analyze the accuracy of registration according to the interval, cubic spline interpolation was applied to the existing point cloud. Additionally, irregular intervals of the point cloud were regenerated into regular intervals. As a result of applying the regenerated point cloud to the surface registration, the surface registration error was not statistically different from the existing point cloud. However, the target registration error significantly lower (p < 0.01). These results indicate that the intervals of point cloud affect the accuracy of registration, and that point cloud with regular intervals can improve the surface registration accuracy.

A Novel Architecture for Internet of Things Based E-Health Systems

Abstract: The emerging Internet of Things (IoT) technology has revolutionized existing medical devices to act smartly towards creating digital world related to health. IoT endeavors an enormous promise in the area of electronic health (e-health) where existing technologies have been deployed to provide rapid access to patient care. IoT based e-health systems involve significant technologies and connected things to monitor, identify, track, manage and store patients' information for ongoing healthcare. A mechanism is required that assures interoperability between devices for processing and use of resources in an efficient manner for the successful deployment of IoT within e-health. The study has coined a unique idea through an exclusive induction of e-governance layer in the concept and implementation of IoT based e-health systems. Therefore, the study targets to develop a novel architecture for incorporation of IoT in the e-health systems by intensively focusing its different layers i.e., sensing, application, transmission, fog and e-governance for having emphasis to get quick response and secured transmission with improved healthcare services. An exploratory study has been conducted to investigate existing literature regarding cutting-edge developments in the arena of e-health. The proposed IoT based e-health architecture is extremely impressive novelty with much comprehensiveness of the concept with the inclusion of very a crucial but previously unnoticed component of Emergency Medical Services (EMS) having its integration through blockchain technology. Another novelty has been presented with the idea of explicitly defining the e-governance layer that encompasses all layers of the proposed architecture in order to achieve quality healthcare services.

DsUnet: A New Network Structure for Detection and Segmentation of Ultrasound Breast Lesions

Abstract: Breast cancer is one of the leading causes of death among the women worldwide. The clinical medical system urgently needs an accurate and automatic breast segmentation method in order to detect the breast ultrasound lesions. Recently, some studies show that deep learning methods based on fully convolutional network, have demonstrated a competitive performance in breast ultrasound segmentation. However, some features are missed in the Unet in case of down-sampling that results in a low segmentation accuracy. Furthermore, there is a semantic gap between the feature maps of decoder and encoder in Unet, so the simple fusion of high and low level features is not conducive to the semantic classification of pixels. In addition, the poor quality of breast ultrasound also affects the accuracy of image segmentation. To solve these problems, we propose a new end-toend network model called Dense skip Unet (DsUnet), which consists of the Unet backbone, short skip connection and deep supervision. The proposed method can effectively avoid the missing of feature information caused by down-sampling and implement the fusion of multilevel semantic information. We used a new loss function to optimize the DsUnet, which is composed of a binary cross-entropy and dice coefficient. We employed the True Positive Fraction (TPF), False Positives per image (FPs) and F -measure as performance metrics for evaluating various methods. In this paper, we adopted the UDIAT 212 dataset and the experimental results validate that our new approach achieved better performance than other existing methods in detecting and segmenting the ultrasound breast lesions. When we used the DsUnet model and new loss function (binary cross-entropy + dice coefficient), the best performance indexes can be achieved, i.e., 0.87 in TPF, 0.13 in FPs/image and 0.86 in F-measure.

Deep Transfer Learning for Histopathological Diagnosis of Cervical Cancer Using Convolutional Neural Networks with Visualization Schemes

Abstract: This study aimed to propose a deep transfer learning framework for histopathological image analysis by using convolutional neural networks (CNNs) with visualization schemes, and to evaluate its usage for automated and interpretable diagnosis of cervical cancer. First, in order to examine the potential of the transfer learning for classifying cervix histopathological images, we pre-trained three state-of-the-art CNN architectures on large-size natural image datasets and then fine-tuned them on small-size histopathological datasets. Second, we investigated the impact of three learning strategies on classification accuracy. Third, we visualized both the multiple-layer convolutional kernels of CNNs and the regions of interest so as to increase the clinical interpretability of the networks. Our method was evaluated on a database of 4993 cervical histological images (2503 benign and 2490 malignant). The experimental results demonstrated that our method achieved 95.88% sensitivity, 98.93% specificity, 97.42% accuracy, 94.81% Youden's index and 99.71% area under the receiver operating characteristic curve. Our method can reduce the cognitive burden on pathologists for cervical disease classification and improve their diagnostic efficiency and accuracy. It may be potentially used in clinical routine for histopathological diagnosis of cervical cancer.

Burnt Human Skin Segmentation and Depth Classification Using Deep Convolutional Neural Network (DCNN)

Abstract: World Health Organization (WHO) manage health-related statistics all around the world by taking the necessary measures. What could be better for health and what may be the leading causes of deaths, all these statistics are well organized by WHO. Burn Injuries are mostly viewed in middle and low-income countries due to lack of resources, the result may come in the form of deaths by serious injuries caused by burning. Due to the non-accessibility of specialists and burn surgeons, simple and basic health care units situated at tribble areas as well as in small cities are facing the problem to diagnose the burn depths accurately. The primary goals and objectives of this research task are to segment the burnt region of skin from the normal skin and to diagnose the burn depths as per the level of burn. The dataset contains the 600 images of burnt patients and has been taken in a real-time environment from the Allied Burn and Reconstructive Surgery Unit (ABRSU) Faisalabad, Pakistan. Burnt human skin segmentation was carried by the use of Otsu's method and the image feature vector was obtained by using statistical calculations such as mean and median. A classifier Deep Convolutional Neural Network based on deep learning was used to classify the burnt human skin as per the level of burn into different depths. Almost 60 percent of images have been taken to train the classifier and the rest of the 40 percent burnt skin images were used to estimate the average accuracy of the classifier. The average accuracy of the DCNN classifier was noted as 83.4 percent and these are the best results yet. By the obtained results of this research task, young physicians and practitioners may be able to diagnose the burn depths and start the proper medication.

Brain Image Enhancement Approach Based on Singular Value Decomposition in Nonsubsampled Shearlet Transform Domain

Abstract: In this work, a novel image enhancement algorithm using NSST and SVD is proposed to improve the definition of the acquired brain images. The input brain image is computed by CLAHE, then the processed brain image and input brain image are decomposed into low- and high-frequency components by NSST, the singular value matrix of the low-frequency component is estimated. The final enhancement image is obtained by inverse NSST. Results of this experiment demonstrate that the proposed technique has good performance in terms of brain image enhancement when compared to other methods.

An Embedded Feature Selection and Multi-Class Classification Method for Detection of the Progression from Mild Cognitive Impairment to Alzheimer's Disease

Abstract: Background: Mild cognitive impairment (MCI) patients are a high-risk group for Alzheimer's disease (AD). Each year, the diagnosed of 10–15% of MCI patients are converted to AD (MCI converters, MCI_C), while some MCI patients remain relatively stable, and unconverted (MCI stable, MCI_S). MCI patients are considered the most suitable population for early intervention treatment for dementia, and magnetic resonance imaging (MRI) is clinically the most recommended means of imaging examination. Therefore, using MRI image features to reliably predict the conversion from MCI to AD can help physicians carry out an effective treatment plan for patients in advance so to prevent or slow down the development of dementia. Methods: We proposed an embedded feature selection method based on the least squares loss function and within-class scatter to select the optimal feature subset. The optimal subsets of features were used for binary classification (AD, MCI_C, MCI_S, normal control (NC) in pairs) based on a support vector machine (SVM), and the optimal 3-class features were used for 3-class classification (AD, MCI_C, MCI_S, NC in triples) based on one-versus-one SVMs (OVOSVMs). To ensure the insensitivity of the results to the random train/test division, a 10-fold cross-validation has been repeated for each classification. Results: Using our method for feature selection, only 7 features were selected from the original 90 features. With using the optimal subset in the SVM, we classified MCI_C from MCI_S with an accuracy, sensitivity, and specificity of 71.17%, 68.33% and 73.97%, respectively. In comparison, in the 3-class classification (AD vs. MCI_C vs. MCI_S) with OVOSVMs, our method selected 24 features, and the classification accuracy was 81.9%. The feature selection results were verified to be identical to the conclusions of the clinical diagnosis. Our feature selection method achieved the best performance, comparing with the existing methods using lasso and fused lasso for feature selection. Conclusion: The results of this study demonstrate the potential of the proposed approach for predicting the conversion from MCI to AD by identifying the affected brain regions undergoing this conversion.

Smart Home Healthcare System Based on Middleware and Counter Neural Network

Abstract: In recent years, with the upsurge of the smart city construction, smart communities and smart medical services have also witnessed rapid development. The development of smart home has achieved remarkable results. The new term of smart home medical treatment is also accompanied by intelligent home into people’s lives. Smart home health is to improve the smart home to bring new experiences of the public life, combined with the wisdom of the medical under another kind of human considerations. The emergence of smart medical treatment is accompanied by smart home and in an attempt to solve the current imbalance in the relationship between doctors and also patients in China, especially the common medical sanitation in hospitals with high sanction, and the heavy workload of hospitals and the contradiction between doctor-patient relationships. Under this general background, this paper proposes smart home healthcare system based on the middleware and counter neural network model. The simulation proves the efficiency.

A Determination Scheme for Quasi-Identifiers Using Uniqueness and Influence for De-Identification of Clinical Data

Abstract: Objectives; The accumulation and usefulness of clinical data have increased with IT development. While using clinical data that needs to be identifiable to obtain meaningful information, it is essential to ensure that data is de-identified and unnecessary clinical information is minimized to protect personal information. This process requires criteria and an appropriate method as there are clear identifiers as well as quasi-identifiers that are not readily identifiable. Methods; To formulate such a method, first, primary quasi-identifiers were selected by classifying information in 20 clinical personal information database tables into Direct-Identifier (DID), Quasi-Identifier (QI), Sensitive Attribute (SA), and Non-Sensitive Attribute (NSA) according to its type. Secondary QIs were then selected by assessing the risk for outliers by measuring uniqueness values of the selected data and scoring re-identification by calculating equivalence class of the influence on other data on QI removal. Third, the risk of re-identification of data users was numeralized and classified. Lastly, the final QI according to user class was determined by comparing the calculated re-identification scores to the threshold values of user classes. Results; Eventually, final QIs ranging from a minimum of 18 to a maximum of 28 were selected by making an assumption about user classes and using it as criteria. Conclusions; The QI selection method presented by the current investigators can be used by researchers at the final checkup stage before they de-identify the selected QIs. Therefore, clinical data users can securely and efficiently use clinical data containing personal information by objectively selecting QIs using the method proposed in the present study.