We evaluate the performance of a deep learning framework for segmenting abdominal fat and muscle using multi-contrast Dixon magnetic resonance (MR) and computed tomography (CT) images. We aim to compare MR image segmentation by testing Dixon fat-only, water-only, and combination of both types of images and comparing the results with CT images. Nineteen subjects underwent abdominal CT and Dixon MR imaging on the same day. For each participant, three pairs of matched axial images from both CT and MR were selected at the intervertebral levels of L2-L3, L3-L4, and L4-L5 for analysis. References labels were generated through semi-automated segmentation of subcutaneous adipose tissue, visceral adipose tissue, and muscle areas. They were then used to train and evaluate a U-Net based Convolutional Neural Network (CNN) framework with a 3-fold cross-validation to compare the segmentation performance across CT, Dixon fat-only and water-only MR images. Combining the fat-only and water-only MR image inputs produced superior results in all labels. Our study demonstrates that CNN-based segmentation performance for abdominal fat and muscle improves with the inclusion of additional input channels, such as combining Dixon fat-only and water-only MR images. While CT results represent the gold standard in abdominal image segmentation, increasing the number of input image channels used in MR segmentation can approach, and even match, the results of CT.

Cardiovascular disease, the leading cause of death in the U.S., affects both the heart and blood vessels. Contrast-enhanced cardiac computed tomography angiography (CTA) is a prominent imaging modality for assessing heart and coronary artery morphology, aiding in the diagnosis of cardiovascular disease. Cine CTA, which captures multiple 3D heart images throughout the cardiac cycle, is particularly useful for evaluating nonischemic cardiomyopathy, a common cause of heart failure unrelated to coronary artery disease. Key biomarkers, such as left ventricular ejection fraction, which measures the amount of blood pumped from the heart’s lower chambers per contraction, and myocardial strain, which evaluates the deformation of the heart muscle during contraction and relaxation, are vital for assessing heart function. We developed an automated artificial intelligence (AI) framework to segment various cardiac structures across all image volumes representing different phases of the cardiac cycle. The framework also automatically extracts three short-axis slices to calculate the myocardial strain in the left ventricle. The results are compared with strain measurements obtained from cardiovascular magnetic resonance (CMR) cine imaging. Our study shows both radial and circumferential strains from cine CTA are comparable to those from cine CMR. The proposed AI framework facilitates a comprehensive evaluation of myocardial function throughout the cine CTA, supporting improved diagnostic accuracy.

Post-stroke rehabilitation of upper extremity (UE) motor function is essential. Despite the widespread use of UE rehabilitation in clinical settings, assessing the success of these treatments is challenging. Methods for evaluating UE motor function include task performance under clinical supervision, patient self-reports, and data analysis from wearable devices equipped with accelerometers and gyroscopes. In prior research, we demonstrated that machine learning and deep learning models using data from a single wrist-worn accelerometer sensor could accurately differentiate between functional and non-functional UE movements in stroke patients. To overcome the limitations of single wrist-worn accelerometer sensors – challenges in capturing the full context of functional movements – this study presents a new deep learning (DL) based framework designed to classify functional and non-functional arm movements in videos captured from a conventional camera. The system is entirely automated and comprises two DL networks. The first network performs human pose estimation, extracting 2D pose key points of the paretic arm(s) and torso on 2-second sequences of the frames. The second network then uses these 2D pose landmarks to classify the movement sequence as functional or non-functional. This system offers two key benefits for rehabilitation. First, it can automatically generate initial annotations for video frames, significantly reducing the time needed for manual labeling. Second, analyzing the effectiveness of UE rehabilitation through third-person videos allows for objective outcome measurement of UE treatments in stroke survivors' home environments.

Cardiac arrhythmias, characterized by irregular heart rhythms, pose a considerable challenge in medical diagnostics due to their diverse and subtle manifestations. Traditional machine learning models, while effective in detecting arrhythmias from electrocardiograms (ECGs), often struggle with these nuances. This study proposes a method that combines sophisticated feature engineering and hierarchical classification to enhance arrhythmia classification. Utilizing a comprehensive ECG database from Chapman University and Shaoxing People’s Hospital, we analyzed the data from 10,646 patients, encompassing 11 types of arrhythmic rhythms along with additional cardiovascular conditions. Data preprocessing involved outlier removal, feature encoding, and the introduction of novel features such as the disparity between atrial and ventricular rates and the Q wave ratio. These new features were crucial in discerning heart rhythm discrepancies. To achieve robust classification, the study employed Random Forest and XGBoost algorithms. A two-tier hierarchical classification model was introduced, initially classifying cardiac rhythms into classes and another model delving into finer distinctions among specific classes. This approach was further enhanced by several oversampling techniques to address varied data distributions across different classes. The results revealed notable improvements in model performance with the integration of new features, yielding increased accuracy and f1-scores. The research, while achieving significant advancements in arrhythmia classification, highlighted the need for further studies with more diverse datasets for broader applicability. The study demonstrates the potential of integrating machine learning with domain-specific features, suggesting a promising direction for advanced diagnostic tools in cardiology, with the potential to improve patient outcomes in arrhythmia treatment.

The rehabilitation of the upper extremity (UE) plays a pivotal role in the recovery process for stroke survivors. Despite the routine practice of UE rehabilitation, assessing its effectiveness poses a significant challenge. Evaluation methods vary, ranging from clinical assessments of UE motor skills and patient feedback to the utilization of wearable technology equipped with accelerometer or gyro sensors. In this study, we employed deep learning approaches such as Multi-Layer Perceptron, Long Short-term Memory, and Gated Recurrent Units, in conjunction with data from accelerometry sensors embedded in wrist-wearable Inertial Measurement Units. Our findings demonstrate the successful differentiation between functional and non-functional UE movements in individuals recovering from strokes with 90% and 74% of classification accuracy for intra-subject and inter-subject models, respectively. The methodologies employed in this study exhibit consistency and reliability, providing precise outcomes for individual and acceptable outcomes for cross-participant evaluations. Notably, our advanced models directly utilized raw accelerometry data instead of manual extracted features that was often used in traditional machine learning algorithms. These innovative techniques offer cost-effective and adaptable tools for monitoring UE functionality in real-world settings. The insights derived from this study hold the potential to be transformative in tailoring rehabilitation strategies for adults affected by strokes.

Humans react both cognitively and emotionally to the built environment. With the recent advent of wearable technology such as mobile EEG devices and biometric wrist monitors that record the electrical activity of the brain and measure physiological data, we have new tools to inspect the internal states of building visitors and to determine whether a measurable, physiological response accompanies their experiences. These measures can shed light on the neural correlates of aesthetic architectural appreciation through specific electrographic signatures. They also create a promising foundation for future architectural design.In this study, we evaluated the cognitive impacts of two distinct architectural conditions—one secular and one religious—on individuals with committed religious faith. Employing machine learning algorithms, specifically the random forest and support vector machine, to analyze the EEG data, we constructed intrasubject and intersubject models to predict whether a given second of EEG data was recorded in the secular or religious building. The best intrasubject model demonstrates just under 80% accuracy on average, reaching up to 90% accuracy for some subjects, indicating a consistent and measurable response within each participant. However, the individual variability in EEG measurements poses challenges for constructing an intersubject model with a sample size of 32 subjects. To achieve success across a broader range of subjects, a larger sample size or additional features may be necessary for developing an effective classifier.

3D sensors such as LiDAR, stereo cameras, and radar have been used in many applications, for instance, virtual or augmented reality, real-time immersive communications, and autonomous driving systems. The output of 3D sensors is often represented in the form of point clouds. However, the massive amount of point cloud data generated from 3D sensors poses big challenges in data storage and transmission. Therefore, effective compression schemes are needed for reducing the bandwidth of wireless networks or storage space of 3D point cloud data. Several point cloud compression (PCC) algorithms have been proposed using signal processing or neural network techniques. In this study, we investigate four state-of-the-art PCC methods using two different datasets with various configurations. The objective of this study is to provide a comprehensive understanding of various approaches in PCC. The results of this paper will be helpful in developing an adaptive 3D point cloud stream compression benchmark that is efficient and benefited from different PCC techniques.

Alzheimer's disease (AD) is a neurodegenerative disease and the most common cause of dementia in older adults. With no known cures, there is a pressing need to find behavioral tasks and biomarkers that can accurately assess and/or predict disease progression in asymptotic patients, as treatment is likely to be most effective at an early stage of AD. On the other hand, artificial intelligence systems are powerful and critical tools to support early detection and diagnosis, treatment, as well as outcome prediction and prognosis evaluation in healthcare. In this study, standard neuropsychological tests and a simple 5.5-minute cognitive task were administered to patients with mild AD or mild cognitive impairment (MCI) (AD group, n=28) and cognitively normal older adults (Control group, n=50). Patients with mild AD or MCI were collapsed together as the AD group. Four different machine learning algorithms were applied to classify patients from healthy controls using the data collected from neuropsychological tests, or the cognitive task, or both. The results of the study revealed that machine learning technique has the potential to assist AD diagnosis using the neuropsychological data, and when combining the neuropsychological and cognitive data, the classification accuracy can be further improved.

Mice are routinely used as an animal model for brain research. Therefore, automatic and robust mouse brain segmentation is an essential task in many applications since it affects the outcomes of the entire analysis. Automated human brain segmentation have been well studied. However, applying existing methods for human brain segmentation directly to the mouse brain is not immediately applicable due to the difference in size, shape and structure between the human and mouse brains. In this paper, we present an automatic mouse brain segmentation method based on a deep convolutional neural network (CNN) called U-Net. Quantitative assessment of the proposed method is performed on 26 mouse brain diffusion MRI studies with a reference standard obtained from expert manual segmentation. We also compared the segmentation result with several state-of-art human brain segmentation methods. The result shows that the proposed CNN model outperforms other methods yielding an average Dice coefficient of 0.974, Hausdorff distance of 3.875 mm, and mean surface distance of 0.134 mm.

The majority of stroke survivors suffer from residual functional deficits in the Upper Extremity (UE) that limits the patients' ability to incorporate the hemiparetic UE into daily function such as reaching and grasping objects. Correctly assessing the degree to which individuals with stroke use their UE would be important for rehabilitation. However, previous work shows that many stroke patients, when observed in the laboratory, can appear to use the paretic arm with adequate ability, yet do not use the arm at home with the expected regularity. The main purposes of this study are to use various machine learning techniques to examine associations between laboratory-based measures of limb functioning and actual home use. The UE kinematics during reaching and grasping tasks were measured in the laboratory, while the use of the paretic arm in the home environment was assessed with portable accelerometry and the Motor Activity Log (MAL). The study identified a subset of the biomechanical features that predicted well the paretic arm use at home. Arm use was also well predicted with clinical evaluations of impairment using Fugl-Meyer and Action Research Arm Test (ARAT) scores. However, combining the clinical and a subset of biomechanical features yielded the best model to predict UE use in individuals with stroke.

Upper Extremity (UE) rehabilitation is often needed post-stroke. The main goal of UE treatment in stroke survivors is to increase the use of the affected UE in the home and community. However, the effectiveness of UE treatments are difficult to quantify because no objective evaluation of UE use exists. In practice, a clinician rates the patient's ability to perform specific motor tasks associated with functional use in a clinic or the patient self-reports the amount or quality of arm movement for a standard set of activities. Both methods do not objectively measure the performance of the affected UE in the home or community environment, and there is growing evidence that motor performance in the laboratory is a poor proxy for the actual amount of UE use. Using a single wrist-worn sensor (i.e., accelerometry data) and machine learning, we have reported that it is possible to separate UE functional use from nonfunctional movement after stroke. Specifically, we reported that we correctly classified sensor data with an average of 94.80% in controls and 88.38% in stroke subjects in intra-subject test trials, and 91.53% for controls and 70.18% in stroke subjects in inter-subject test trials. In this paper, we employed feature selection techniques and explored different machine learning methods to improve the classification accuracy. Our enhanced methods are robust and reliable, and work in both intra-subject and inter-subject training and testing. Our result showed better accuracy in stroke patients than previously reported with the same dataset. The enhanced models reached an average of 96% accuracy in control subjects and 94% in stroke subjects for intra-subject trials, and an average of 90% accuracy in control subjects and 83% in stroke subjects for the inter-subject trials. The proposed methods provide an inexpensive and feasible way to quantify the UE functional use in home and community. This information can provide guidance for clinical practice in the rehabilitative care of adults recovering from stroke.

Modern multidimensional medical imaging technology produces very large amount of data especially from the computed tomography modality. These volumetric dataset opens new demands for big data storage and high-speed communication systems which may be alleviated by image compression techniques. Current image compression schemes adopted in the DICOM standard do not exploit the inter-slice correlation within three-dimensional (3D) dataset. Video compression may have a potential to improve the compression ratio by reducing the redundancy in volumetric medical images. In this paper, we compare the performance of five lossless video codecs (H264, H265, Lagarith, MSU, MLC) and three still-image codecs (JPEG, JPEG2000, JPEG-LS) using 3D medical computed tomography datasets. Performance evaluation shows that video codecs improve the compression ratio compared to JPEG and JPEG2000 while being competitive to JPEG-LS.

Data and methodology sharing is essential for progression of scientific research. Several research groups have built tools for medical big data (MBD) processing applicable to Diffusion Tensor MRI (DTI) processing pipelines. In this paper, we propose a framework enabling methodology sharing (i.e. harmonization) to facilitate the reproducibility in DTI processing.

Big data processing has introduced new ideas in the applications of bacterial analysis in recent years. This paper aims to develop an effective framework to automatically extract quantitative knowledge relating to bacterial motility through processing a sequence of large-scale microscopic images of bacterial movements. It was hypothesized that motile bacteria move according to a conceptual model referred to as the “run and tumble” modes. The tremendous amount of microscopic image sequences contains abundant bacteria motility information. However, the traditional experimental methods to identify the trajectories, and segment them to “run” and “tumble” modes are time consuming, subjective, and are not suitable for large-amount and large-scale data analysis. The proposed framework enables rapid and reliable bacteria motility analysis which is important in pathogenic risk assessment, water treatment, soil and groundwater bio-remediation, etc. Furthermore, the framework can be easily adapted to study the diversity and habitats of bacteria from other experimental microscopic images.

E-commerce plays a key role in business success nowadays. Therefore, the performance of E-commerce websites is critical. E-commerce websites generate a large amount of data that is often used for performance evaluation. Many website evaluation methods have been proposed, but the social media factor is usually not taken into consideration. In this paper, Twitter data is utilized for big data analytics and several Twitter performance indexes are proposed to assist the website performance evaluation. The result is compared with the performance evaluation result using several commercial tools for 13 selected E-commerce websites in Saudi Arabia.