Expert-level segmentation using deep learning for volumetry of polycystic kidney and liver.

PURPOSE: Volumetry is used in polycystic kidney and liver diseases (PKLDs), including autosomal dominant polycystic kidney disease (ADPKD), to assess disease progression and drug efficiency. However, since no rapid and accurate method for volumetry has been developed, volumetry has not yet been established in clinical practice, hindering the development of therapies for PKLD. This study presents an artificial intelligence (AI)-based volumetry method for PKLD. MATERIALS AND METHODS: The performance of AI was first evaluated in comparison with ground-truth (GT). We trained a V-net-based convolutional neural network on 175 ADPKD computed tomography (CT) segmentations, which served as the GT and were agreed upon by 3 experts using images from 214 patients analyzed with volumetry. The dice similarity coefficient (DSC), interobserver correlation coefficient (ICC), and Bland-Altman plots of 39 GT and AI segmentations in the validation set were compared. Next, the performance of AI on the segmentation of 50 random CT images was compared with that of 11 PKLD specialists based on the resulting DSC and ICC. RESULTS: The DSC and ICC of the AI were 0.961 and 0.999729, respectively. The error rate was within 3% for approximately 95% of the CT scans (error<1%, 46.2%; 1%≤error<3%, 48.7%). Compared with the specialists, AI showed moderate performance. Furthermore, an outlier in our results confirmed that even PKLD specialists can make mistakes in volumetry. CONCLUSIONS: PKLD volumetry using AI was fast and accurate. AI performed comparably to human specialists, suggesting its use may be practical in clinical settings.

Evaluation of Fear Using Nonintrusive Measurement of Multimodal Sensors.

Most previous research into emotion recognition used either a single modality or multiple modalities of physiological signal. However, the former method allows for limited enhancement of accuracy, and the latter has the disadvantages that its performance can be affected by head or body movements. Further, the latter causes inconvenience to the user due to the sensors attached to the body. Among various emotions, the accurate evaluation of fear is crucial in many applications, such as criminal psychology, intelligent surveillance systems and the objective evaluation of horror movies. Therefore, we propose a new method for evaluating fear based on nonintrusive measurements obtained using multiple sensors. Experimental results based on the t-test, the effect size and the sum of all of the correlation values with other modalities showed that facial temperature and subjective evaluation are more reliable than electroencephalogram (EEG) and eye blinking rate for the evaluation of fear.

A Fuzzy-Based Fusion Method of Multimodal Sensor-Based Measurements for the Quantitative Evaluation of Eye Fatigue on 3D Displays.

With the rapid increase of 3-dimensional (3D) content, considerable research related to the 3D human factor has been undertaken for quantitatively evaluating visual discomfort, including eye fatigue and dizziness, caused by viewing 3D content. Various modalities such as electroencephalograms (EEGs), biomedical signals, and eye responses have been investigated. However, the majority of the previous research has analyzed each modality separately to measure user eye fatigue. This cannot guarantee the credibility of the resulting eye fatigue evaluations. Therefore, we propose a new method for quantitatively evaluating eye fatigue related to 3D content by combining multimodal measurements. This research is novel for the following four reasons: first, for the evaluation of eye fatigue with high credibility on 3D displays, a fuzzy-based fusion method (FBFM) is proposed based on the multimodalities of EEG signals, eye blinking rate (BR), facial temperature (FT), and subjective evaluation (SE); second, to measure a more accurate variation of eye fatigue (before and after watching a 3D display), we obtain the quality scores of EEG signals, eye BR, FT and SE; third, for combining the values of the four modalities we obtain the optimal weights of the EEG signals BR, FT and SE using a fuzzy system based on quality scores; fourth, the quantitative level of the variation of eye fatigue is finally obtained using the weighted sum of the values measured by the four modalities. Experimental results confirm that the effectiveness of the proposed FBFM is greater than other conventional multimodal measurements. Moreover, the credibility of the variations of the eye fatigue using the FBFM before and after watching the 3D display is proven using a t-test and descriptive statistical analysis using effect size.

Robust pedestrian detection by combining visible and thermal infrared cameras.

With the development of intelligent surveillance systems, the need for accurate detection of pedestrians by cameras has increased. However, most of the previous studies use a single camera system, either a visible light or thermal camera, and their performances are affected by various factors such as shadow, illumination change, occlusion, and higher background temperatures. To overcome these problems, we propose a new method of detecting pedestrians using a dual camera system that combines visible light and thermal cameras, which are robust in various outdoor environments such as mornings, afternoons, night and rainy days. Our research is novel, compared to previous works, in the following four ways: First, we implement the dual camera system where the axes of visible light and thermal cameras are parallel in the horizontal direction. We obtain a geometric transform matrix that represents the relationship between these two camera axes. Second, two background images for visible light and thermal cameras are adaptively updated based on the pixel difference between an input thermal and pre-stored thermal background images. Third, by background subtraction of thermal image considering the temperature characteristics of background and size filtering with morphological operation, the candidates from whole image (CWI) in the thermal image is obtained. The positions of CWI (obtained by background subtraction and the procedures of shadow removal, morphological operation, size filtering, and filtering of the ratio of height to width) in the visible light image are projected on those in the thermal image by using the geometric transform matrix, and the searching regions for pedestrians are defined in the thermal image. Fourth, within these searching regions, the candidates from the searching image region (CSI) of pedestrians in the thermal image are detected. The final areas of pedestrians are located by combining the detected positions of the CWI and CSI of the thermal image based on OR operation. Experimental results showed that the average precision and recall of detecting pedestrians are 98.13% and 88.98%, respectively.

Human detection based on the generation of a background image by using a far-infrared light camera.

The need for computer vision-based human detection has increased in fields, such as security, intelligent surveillance and monitoring systems. However, performance enhancement of human detection based on visible light cameras is limited, because of factors, such as nonuniform illumination, shadows and low external light in the evening and night. Consequently, human detection based on thermal (far-infrared light) cameras has been considered as an alternative. However, its performance is influenced by the factors, such as low image resolution, low contrast and the large noises of thermal images. It is also affected by the high temperature of backgrounds during the day. To solve these problems, we propose a new method for detecting human areas in thermal camera images. Compared to previous works, the proposed research is novel in the following four aspects. One background image is generated by median and average filtering. Additional filtering procedures based on maximum gray level, size filtering and region erasing are applied to remove the human areas from the background image. Secondly, candidate human regions in the input image are located by combining the pixel and edge difference images between the input and background images. The thresholds for the difference images are adaptively determined based on the brightness of the generated background image. Noise components are removed by component labeling, a morphological operation and size filtering. Third, detected areas that may have more than two human regions are merged or separated based on the information in the horizontal and vertical histograms of the detected area. This procedure is adaptively operated based on the brightness of the generated background image. Fourth, a further procedure for the separation and removal of the candidate human regions is performed based on the size and ratio of the height to width information of the candidate regions considering the camera viewing direction and perspective projection. Experimental results with two types of databases confirm that the proposed method outperforms other methods.

Assessment of eye fatigue caused by 3D displays based on multimodal measurements.

With the development of 3D displays, user's eye fatigue has been an important issue when viewing these displays. There have been previous studies conducted on eye fatigue related to 3D display use, however, most of these have employed a limited number of modalities for measurements, such as electroencephalograms (EEGs), biomedical signals, and eye responses. In this paper, we propose a new assessment of eye fatigue related to 3D display use based on multimodal measurements. compared to previous works Our research is novel in the following four ways: first, to enhance the accuracy of assessment of eye fatigue, we measure EEG signals, eye blinking rate (BR), facial temperature (FT), and a subjective evaluation (SE) score before and after a user watches a 3D display; second, in order to accurately measure BR in a manner that is convenient for the user, we implement a remote gaze-tracking system using a high speed (mega-pixel) camera that measures eye blinks of both eyes; thirdly, changes in the FT are measured using a remote thermal camera, which can enhance the measurement of eye fatigue, and fourth, we perform various statistical analyses to evaluate the correlation between the EEG signal, eye BR, FT, and the SE score based on the T-test, correlation matrix, and effect size. Results show that the correlation of the SE with other data (FT, BR, and EEG) is the highest, while those of the FT, BR, and EEG with other data are second, third, and fourth highest, respectively.

Noise reduction in brainwaves by using both EEG signals and frontal viewing camera images.

Electroencephalogram (EEG)-based brain-computer interfaces (BCIs) have been used in various applications, including human-computer interfaces, diagnosis of brain diseases, and measurement of cognitive status. However, EEG signals can be contaminated with noise caused by user's head movements. Therefore, we propose a new method that combines an EEG acquisition device and a frontal viewing camera to isolate and exclude the sections of EEG data containing these noises. This method is novel in the following three ways. First, we compare the accuracies of detecting head movements based on the features of EEG signals in the frequency and time domains and on the motion features of images captured by the frontal viewing camera. Second, the features of EEG signals in the frequency domain and the motion features captured by the frontal viewing camera are selected as optimal ones. The dimension reduction of the features and feature selection are performed using linear discriminant analysis. Third, the combined features are used as inputs to support vector machine (SVM), which improves the accuracy in detecting head movements. The experimental results show that the proposed method can detect head movements with an average error rate of approximately 3.22%, which is smaller than that of other methods.

Enhanced perception of user intention by combining EEG and gaze-tracking for brain-computer interfaces (BCIs).

Speller UI systems tend to be less accurate because of individual variation and the noise of EEG signals. Therefore, we propose a new method to combine the EEG signals and gaze-tracking. This research is novel in the following four aspects. First, two wearable devices are combined to simultaneously measure both the EEG signal and the gaze position. Second, the speller UI system usually has a 6 × 6 matrix of alphanumeric characters, which has disadvantage in that the number of characters is limited to 36. Thus, a 12 × 12 matrix that includes 144 characters is used. Third, in order to reduce the highlighting time of each of the 12 × 12 rows and columns, only the three rows and three columns (which are determined on the basis of the 3 × 3 area centered on the user's gaze position) are highlighted. Fourth, by analyzing the P300 EEG signal that is obtained only when each of the 3 × 3 rows and columns is highlighted, the accuracy of selecting the correct character is enhanced. The experimental results showed that the accuracy of proposed method was higher than the other methods.