Search

Pneumonia is a lung disease caused by bacterial or viral infection. Early diagnosis is an important factor for successful treatment. In this study, we use three well-known convolutional neural network models, namely Faster RCNN ResNet-101, Mask RCNN ResNet-101, and Mask RCNN ResNet-50 for detection of pneumonia. We use data augmentation, transfer learning and fine-tuning in the training stage. Experimental results show that different networks have different characteristics on the same dataset. Therefore, we propose a multiple deep learner approach to improve the prediction performance via combination of different object detection models. As a result, the proposed approach can find more opacity areas of the lungs where the early symptoms are not evident. While maintaining the prediction accuracy, the proposed method can predict the bounding box size more precisely with a higher confidence score.

This paper presents a voting-based system via crowdsourcing for detecting traffic collision. The proposed system, as opposed to existing traffic collision avoidance systems, does not consider collision avoidance problem. Rather, it treats a traffic collision as a detection problem and poses a crowd voting strategy to detect collision and thereby alleviate traffic congestion caused by accidents in the real-world driving condition. Our experiments on traffic collision detection problem validate the significant accuracy of the proposed system.

Deep brain stimulation (DBS) has been found useful for treating neural diseases such as the Parkinson's disease, while the mechanism is not well understood and the DBS is suspected of inducing various side effects. This paper presents a microsystem suitable for studying the mechanism of the DBS. The microsystem contains eight channels of neural recording and stimulation circuits, an analog-to-digital converter, and a digital information hub. In addition, to facilitate long-term study, the microsystem is implantable and batteryless. Both power and data are transmitted wirelessly through a single coil by the time-division-multiple-access (TDMA) technique. The microsystem has been designed and fabricated with the 0.18μm CMOS technology. The pilot testing results are presented and discussed.

The Pakinson's Disease (PD) is one of the most common diseases for the elderly, and is attributed to the degeneration of neurons in the brain. This paper presents a microsystem suitable for recording the neural activity relating to the PD symptons for a long term, as well as for stimulating the neurons in the deep brain. As many studies indicate that changes of low-frequency field potentials correlate with the PD symptoms, eight chopper amplifiers with tunable bandwidth are designed to record not only low-frequency potentials but also neuronal spikes. Two stimulation circuits with adjustable stimulation voltage are also incorporated to investigate how neural activities change with the stimulation condition. Moreover, the microsystem contains wireless modules for both power and data transmission, facilitating long-term implantation. The functionality of microsystem is tested and discussed in this paper.