Session Detail

This study investigated the impact of body mass index (BMI) in prospectively ECG-triggered coronary CT angiography (CCTA) to achieve sufficient and consistent image quality across a diverse patient population. All CCTA scans were performed on a 320-slice multi-detector CT with default protocol settings. Automatic exposure control (AEC) was used in conjunction with iterative reconstruction (IR) to ensure sufficient diagnostic information at the lowest radiation dose. Multiple linear regression methods were used to analyze how tube voltage, tube current and chest circumference varied according to BMI. A total of 1509 consecutive CCTA examina-tions were enrolled in this study (468 women, 1041 men; age range 25-89 years; heart rate range 40.08-80.75 bpm; BMI range 15.76-47.35 kg/m2). The regression model suggested that 100-, 120- and 135-kVp CCTA scans should be used in patients with BMI less than 26, 30 and 32 kg/m2, respectively. As for patients with BMI > 32 kg/m2, although the efficacy of AEC on compensating patient attenuation is limited, the image noise can be reduced by increasing the blending level of IR-FBP.

 

Deep learning methods have become increasingly important as the advancement of computing speed and capacity during recent years. When applied to computer vision and image processing, deep learning has great performance on object classification, object detection, image segmentation, etc. In this paper, we apply convolutional neural network to classify ecological data such as 19 types of bee wings. One difficulty in the ecology classification is the limited amounts of the dataset. Thus, image augmentation method is used to increase the dataset size. The original dataset of 750 images is enlarged to 19,000 images, among them 15,200 images used for training (i.e., 800 images for each class) and 3,800 images for testing. With the augmented images, we train our model to classify the types of bees according to their wings. Furthermore, we apply transfer learning technique to improve model’s classification accuracy. A pre-trained neural network model is applied on the ecological dataset. The test accuracy of 98.34% is achieved after data augmentation. And 93.79% test accuracy for trans-fer learning, which is used to build one-class classification and recognize “unknown” class.

 

A lot of severe disabled patients such as neurodegenerative disease or spinal cord injury are difficult to use traditional communication aids and then cause many problems in their daily lives. To help the severely disabled patients, steady state visually evoked potentials based brain-computer interface (SSVEP-based BCI) is one of the efficient ways to easy the patients in communicating with other people or devices. In this study, a modular deep neural network is developed to integrate the decisions by using different types of features. To effective represent the characteristics of biological signal elicited SSVEP, the canonical correlation analysis, fast Fourier transform, and magnitude-squared coherence are adopt to extract the features. To improve the communication effectives of SSVEP-base BCI, a modular deep neural network (MDNN) is developed to find the feature dependent decisions by using different features and then to fuse these decisions for finding a precise recognition result. The experimental results showed that MDNN produce higher accuracy compare to other approaches. Therefore, the proposed MDNN can effectively help severe disabled patients in interacting with other people or devices.

 

The development of medical technologies is improving medicine practice and healthcare approach. Nevertheless, in developing countries, healthcare is a major issue mainly because of economic difficulties. This research proposes a real time monitoring platform to solve patient data digitalization and integration in health systems in order to improve vital signs data management. This platform is a support for health professionals to follow up and visualize differently patient’s vital signs. The web page of the platform allows staff to connect and check the vital signs wherever they are. The platform is designed as a set of three parts that connect and inter communicate: the medical equipment, the Raspberry Pi based medical gateway and the medical cloud server. By using cheap and reliable technologies, open source software and resources, this low cost and low energy consumption system match with developing countries needs and realities.

 

Optimal approaches in fall risk assessment involve interdisciplinary collaboration of assessment. We hypothesized that the high dimensionality objective sensor based parameters, followed by a feature selection and dimensionality reduction process, be able to discriminate elderly nonspecific fallers. 31 community-living elder who were beyond 60 years old (faller: n=15; non faller: n=16) were recruited. The measurements include gait, balance and ankle proprioception performances. Linear Discriminant Analysis (LDA) and Generalized Discriminant Analysis (GDA) were further applied to obtain more discriminative feature space. Receiver-Operator Characteristic (ROC) curves were constructed to compare the classification quality in all the features. The AUC of ROC was GDA dimensionality reduction feature (1), LDA dimensionali-ty reduction feature (0.99), Proprioception (0.752), IMU (0.745) and COP (0.72), respectively. The experimental result show the GDA feature has the best classification quality and the addi-tional advantage in combination of interdisciplinary multi-factorial fall risk assessment.

 

Neurofeedback training (NFT) of electroencephalogram (EEG) is a psychophysiological procedure that allows users to learn self-regulation of their cortical oscillations. Trained alpha activity after NFT is associated with cognitive performance, its neural origin, however, remains unknown. The present study aimed to explore possible equivalent current dipoles (ECDs) of the trained alpha activity. Thirty-six healthy participants were recruited and randomly assigned to either an Alpha group with feedback of 8-12 Hz (n=20) or a Sham group with feedback of 4-Hz bandwidth amplitude selected randomly from 7-20 Hz (n=16). The Alpha group exhibited progressively significant increase of 8-12-Hz amplitude exclusively. Furthermore, the Alpha group had reliable controllability of the trained alpha occurrence in a block deign. ECDs of the trained alpha activity existed in the posterior cingulate cortex, precuneus, middle temporal cortex, and hippocampal formation, which are associated with several cognitive functions. There was no ECD within the occipital cortex. These results suggest that the trained alpha activity differs from classic alpha rhythm and may play an active role in cognitive performance.

 

In this study, we investigated the crucial portion of the k-space data for the reliable magnetic resonance (MR) temperature image of radiofrequency ablation (RFA). A MR compatible RFA system, including the titanium needle electrode, the brass neutral electrode and the cooper conduction wires, was developed. The k-space data of RFA were obtained with a 3T magnetic resonance imaging (MRI) equipment while the compatibility of needle electrode was confirmed in a 1.5T MRI scanner. The temperature image was derived from the k-space data based on the proton resonance frequency shift technique. The computational complexity of the k-space data for temperature image was reduced by replacing a portion of the data by zero. We found that the temperature image of ablative lesions would be remained reliable with 23.4% of the k-space data.