The effectiveness of domain translation in addressing image-based problems of Unsupervised Domain Adaptation (UDA) depends on the quality of the translated images and the preservation of crucial discriminative features. However, achieving high-quality and stable translations typically requires paired data, which poses a challenge in scenarios with limited annotations in the target domain. To address this issue, this paper proposes a novel method termed Stain-Guided Cycle Diffusion (SGCD), employing a dual diffusion model with bidirectional generative constraints to synthesize highly realistic data for downstream task fine-tuning. The bidirectional generative constraints ensure that the translated images retain the features critical to the downstream model in properly controlling the generation process. Additionally, a stain-guided consistency loss is introduced to enhance the denoising capability of the dual diffusion model, thereby improving the quality of images translated between different domains using latents from one domain and a diffusion model trained on another. Experiments conducted on four public datasets demonstrate that SGCD can effectively enhance the performance of downstream task models on the target domain.

Hyperspectral image classification (HSIC) has received considerable interest in recent years where most techniques are developed to classify images with background (BKG) removed by ground truth (GT). Unfortunately, in real scenarios, obtaining complete BKG knowledge is generally infeasible. Accordingly, HSIC performed with no BKG (HSIC-NB) is not realistic. Most importantly, many techniques claim to work well for HSIC-NB but perform poorly with BKG included. This article investigates issues arising from BKG in HSIC and further presents a new approach to HSIC with BKG (HSIC-B), called one class detection (OCD), which is based on the well-known hyperspectral subpixel detection technique, constrained energy minimization (CEM). In order for OCD to perform multiclass classification, OCD is further extended to hierarchical OCD (HOC) which is particularly designed to classify multiple classes in a hierarchical tree where each layer uses an iterative kernel CEM (IKCEM) or an iterative kernel target-constrained interference-minimized filter (IKTCIMF) to detect one class at a time for classification. Since M classes are classified by OCD in${M} -1$layers in a hierarchical tree, a new concept of class classification priority (CCP) derived from CEM is specifically designed to rank all the classes along the tree in a prioritized order according to their CCP scores. The experimental results demonstrate that hierarchical OCD (HOCD) works well and performs significantly better than many existing HSIC-NB methods at the expense of slightly reduced classification accuracy compared to HSIC-N methods.

Hepatic steatosis is one of the most obvious indicators of nonalcoholic fatty liver disease. However, the presence of many regions with a similar color and shape as lipid droplets in the histopathological image complicates the task of detecting genuine lipid droplets using automated methods. Accordingly, the present study proposes a boundary segmentation and matching (BSMatch) algorithm for the segmentation of lipid droplets based on their unique boundary characteristics. A two-branch RnB-Unet model is trained to segment the regions and boundaries of the droplets, respectively, in accordance with a boundary matching (BM) loss which enforces the consistency between them. A boundary matching score (BMS) measure is then used to improve the precision of the instance segmentation evaluation process by discarding segmented regions which are not well-matched with their predicted boundaries. The experimental results obtained using a H&E-stained liver slide dataset show that BSMatch outperforms existing methods in terms of both the IoU and the F1-score. The BSMatch results are used to predict the fat percentage in hepatocytes (FPH) in liver whole slide images. The predicted FPH values are well correlated with the steatosis grades assigned by experienced pathologists. Thus, BSMatch appears to have significant promise for NAFLD diagnosis in clinical contexts.

Hepatocyte nucleus segmentation in histopathology images is vital for diagnostics. However, varying slide background due to cutting and staining poses a great challenge for domain-agnostic segmentation, which limits model generalization. This paper first proposes an anti-background perturbation consistency (APC) loss to mitigate the influence of image background on model decisions and controlled background perturbations to enhance generalizability, while still maintaining feature consistency. Since convolutional neural networks (CNNs) often prioritize local texture over global shape which also limits generalization, we then introduce the concept of local self-information into the texture probability (TP) loss to reduce over-focus of CNN on local textures. To avoid model convergence to saddle points during training which yields varying outcomes and unstable performance, we finally conclude with a meta-learner which combines results from multiple models to improve stability and better decision-making. At the end, our developed APC coupled with the texture reduction ensemble model (TREM) effectively increases model generalizability across diverse data without fine-tuning model parameters. The code of this study are available at https://github.com/s07362022/ATE.

Anomaly detection (AD) requires spectral and spatial information to differentiate anomalies from their surrounding data samples. To capture spatial information, a general approach is to utilize local windows in various forms to adapt local characteristics of the background (BKG) from which unknown anomalies can be detected. This article develops a new approach, called iterative spectral–spatial hyperspectral AD (ISSHAD), which can improve an anomaly detector in its performance via an iterative process. Its key idea is to include an iterative process that captures spectral and spatial information from AD maps (ADMaps) obtained in previous iterations and feeds these anomaly maps back to the current data cube to create a new data cube for the next iteration. To terminate the iterative process, a Tanimoto index (TI)-based automatic stopping rule is particularly designed. Three types of spectral and spatial information, ADMaps, foreground map (FGMap), and spatial filtered map (SFMap), are introduced to develop seven various versions of ISSHAD. To demonstrate its full utilization in improving AD performance, a large number of extensive experiments are performed for ISSHAD along with its detailed comprehensive analysis among several most recently developed anomaly detectors, including classic, dual-window-based, low-rank representation model-based, and tensor-based AD methods for validation.

One-class classification aims to learn one-class models from only in-class training samples. Because of lacking out-of-class samples during training, most conventional deep learning based methods suffer from the feature collapse problem. In contrast, contrastive learning based methods can learn features from only in-class samples but are hard to be end-to-end trained with one-class models. To address the aforementioned problems, we propose alternating direction method of multipliers based sparse representation network (ADMM-SRNet). ADMM-SRNet contains the heterogeneous contrastive feature (HCF) network and the sparse dictionary (SD) network. The HCF network learns in-class heterogeneous contrastive features by using contrastive learning with heterogeneous augmentations. Then, the SD network models the distributions of the in-class training samples by using dictionaries computed based on ADMM. By coupling the HCF network, SD network and the proposed loss functions, our method can effectively learn discriminative features and one-class models of the in-class training samples in an end-to-end trainable manner. Experimental results show that the proposed method outperforms state-of-the-art methods on CIFAR-10, CIFAR-100 and ImageNet-30 datasets under one-class classification settings. Code is available at https://github.com/nchucvml/ADMM-SRNet.

Many driver monitoring systems (DMSs) have been proposed to reduce the risk of human-caused accidents. Traditional DMSs focus on detecting specific predefined abnormal driving behaviors, such as drowsiness or distracted driving, using generic models trained with the data collected during abnormal driving. However, it is difficult to collect sufficient representative training data to construct generic detection models, which are applicable to all drivers. Consequently, this paper proposes a new personal-based hierarchical DMS (HDMS). During driving, the first layer of the proposed HDMS detects normal and abnormal driving behavior based on normal personal driving models represented by sparse representations. When abnormal driving behavior is detected, the second layer of the HDMS further determines whether the behavior is drowsy driving behavior or distracted driving behavior. The experimental results obtained for three datasets show that the proposed HDMS outperforms existing state-of-the-art DMS methods in detecting normal driving behavior, drowsy driving behavior, and distracted driving behavior.

In this paper, we propose a novel annotation placement approach for revealing information about foreground objects in surveillance videos. To arrange positions of annotations, spatiotemporal coherence between annotations and foreground objects is applied. The annotation placement problem is formulated as an optimization problem with respect to spatiotemporal coherence of annotations and foreground objects. The optimization problem is effectively solved using Markov random fields. To the best of our knowledge, this paper is the first work that discusses and solves the annotation placement problem for surveillance videos by considering the relationships between annotations and foreground objects with trajectories. As shown in the experiments, the proposed approach can arrange annotations based on the moving trajectories of foreground objects and prevent the occlusions between different annotations and foreground objects. It also achieves better quantitative and qualitative results compared with state-of-the-art approaches.

Recent big data analysis usually involves datasets with features collected from various sources, where each feature may have different importance, and the training datasets may not be uniformly sampled. To improve the prediction quality of realworld learning problems, we propose a local radial basis function network that is capable of handling both nonuniform sampling density and heterogeneous features. Nonuniform sampling is resolved by estimating local sampling density and adjust the width of the Gaussian kernels accordingly, and heterogeneous features are handled by scaling each dimension of the feature space asymmetrically. To make the learner aware of inter-feature relationship, we propose a feature importance optimization technique base on L-BFGS-B algorithm, using the leave-one-out cross-validation mean squared error as the objective function. Leave-one-out cross-validation used to be a very time consuming process, but the optimization has been made practical by the fast cross-validation capability of local RBFN. Our experiments show that when both nonuniform sampling density and interfeature relationship are properly handled, a simple RBFN can outperform more complex kernel-based learning models such as support vector regressor on both mean-squared-error and training speed.

The similarity comparisons between single-task walking and dual-task walking on Alzheimer's disease (AD) patients has been commonly performed for cognitive declination measurement. This paper presents a personalized gait similarity measurement approach based on Hidden Markov model for the self-comparison between the single-task walking and dual-task walking. Compared with traditional approaches which use statistics parameters comparison on normal group and AD group, the proposed personalized HMM-based self-comparison approach can avoid the dilemma resulted from personal differences such as walking habits and physical conditions such as height and weight. In this paper, two groups, 42 AD patients and 64 healthy control (HC) people, participate the experiments. The results show the promising of the proposed approach in comparing the AD from the normal people.

Trajectories provide spatial-temporal information of foreground objects for event clustering and analysis. Because of the kinematic properties of foreground objects, the lengths of trajectories will be different which lead to the length problem of assessing similarity between two or more trajectories. To solve the problem, we propose a novel descriptor named trajectory kinematics descriptor to represent trajectories based on the kinematic properties from the point-of-view of Frenet-Serret frames. As shown in the experiments, applying the proposed trajectory kinematics descriptor for event clustering can achieve better F-measure scores compared to the state-of-the-art methods.

Time-up and Go Test (TUGT) is a common clinical motility test for the elderly. Currently it is measured manually, which is human resource consuming, less accurate, and limited in capability of measuring sophisticated parameters. In this paper, we propose a system consisting of three inertial-sensor-based measuring components, a remote controller, and an analysis software for automatic balance and gait analysis in TUGT. This system is designed to be of easy use with the software automatically measuring the time intervals of stand-up time, sit-down time, and the total time of test. In addition, the system can also obtain the gait information of stride time, swing time, and stance time. To validate the system, we compared the performance with that obtained from stopwatch using videos as the gold standard. The results indicate that the system not only provides a convenient way of measurement but also achieves higher accurate and reliable measurement results than using stopwatch. We use this system to compare the Alzheimer's disease patients (ADs) with health controllers (HCs). The results show that the ADs need longer stand-up time, sit-down time, and total time while taking the TUGT compared to the HCs. The stride time, swing time, and stance time are also longer than the HCs'. Such a friendly system provides high potential in tracking long-time patterns for diagnosis.

Alzheimer's disease (AD) is becoming one of the major diseases of the elderly. Traditionally, patients take questionnaires or do some balance tests for clinical evaluation. However, results with such evaluation are subjective. For more objective quantitative measurement, this paper uses an inertial-sensor-based device to measure the gait information while participants walking. In the experiment, the participants are asked to walk on a 40m strike line and take single-task and dual-task tests. In the dual-task test, the participants are asked to count down from 100. This paper presents a stride detection algorithm to automatically acquire gait information of each gait cycle from the acceleration and angular velocity signals. Features are calculated from those inertial signals. After feature generation, we do feature selection to select the significant feature. Then, a probabilistic neural networks (PNNs) is used to classify if the participants suffer from AD. In this paper, we provide an objective way to evaluate the situation of the participants. The experimental results successfully validate the effectiveness of the proposed device and the proposed algorithm with an overall classification accuracy rates are 63.33% and 70.00% in women and men group, respectively.

Trajectories of foreground objects provide rich and continuous motion information for event analysis in surveillance videos. These trajectories can be obtained by tracking foreground objects frame by frame. When appearances of foreground objects significantly change during occlusions, tracking results may become incorrect. To solve the occlusion problem, we propose a real-time post processing method based on trajectory links. By analyzing trajectory links, our method can retrieve individual trajectories of occluded foreground objects. As shown in the experiments, our method can achieve better performance compared to previous tracking methods.

Despite the fact that the concept of using long-term personal records has been well recognized for decades, sparse patient physical exam records are still the main base for disease diagnosis and treatment. One of the main reasons for this is the lack of devices and systems for effectively collecting real-time personal data and turning the data into health-meaningful information.

To reduce human efforts in browsing long surveillance videos, synopsis videos are proposed. Traditional synopsis video generation applying optimization on video tubes is very time consuming and infeasible for real-time online generation. This dilemma significantly reduces the feasibility of synopsis video generation in practical situations. To solve this problem, the synopsis video generation problem is formulated as a maximum a posteriori probability (MAP) estimation problem in this paper, where the positions and appearing frames of video objects are chronologically rearranged in real time without the need to know their complete trajectories. Moreover, a synopsis table is employed with MAP estimation to decide the temporal locations of the incoming foreground objects in the synopsis video without needing an optimization procedure. As a result, the computational complexity of the proposed video synopsis generation method can be significantly reduced. Furthermore, as it does not require prescreening the entire video, this approach can be applied on online streaming videos.

Despite patients with Alzheimer's disease (AD) were reported of revealing gait disorders and balance problems, there is still lack of objective quantitative measurement of gait patterns and balance capability of AD patients. Based on an inertial-sensor-based wearable device, this paper develops gait and balance analyzing algorithms to obtain quantitative measurements and explores the essential indicators from the measurements for AD diagnosis. The gait analyzing algorithm is composed of stride detection followed by gait cycle decomposition so that gait parameters are developed from the decomposed gait details. On the other hand, the balance is measured by the sway speed in anterior-posterior (AP) and medial-lateral (ML) directions of the projection path of body's center of mass (COM). These devised gait and balance parameters were explored on twenty-one AD patients and fifty healthy controls (HCs). Special evaluation procedure including single-task and dual-task walking experiments for observing the cognitive function and attention is also devised for the comparison of AD and HC groups. Experimental results show that the wearable instrument with the designed gait and balance analyzing system is a promising tool for automatically analyzing gait information and balance ability, serving as assistant indicators for early diagnosis of AD.

In this paper, we propose a novel orientation-aware Urquhart graph based spatial face context representation method to efficiently describe the spatial relationship among faces in group photos. We combine graph matching with orientations of graph edges to assess the similarity of spatial face contexts from different group photos. The experimental results show that our method can find more structurally similar group photos compared to the state-of-the-art spatial face context representation methods.

The disparity estimation problem is commonly solved using graph cut (GC) methods, in which the disparity assignment problem is transformed to one of minimizing global energy function. Although such an approach yields an accurate disparity map, the computational cost is relatively high. Accordingly, this paper proposes a hierarchical bilateral disparity structure (HBDS) algorithm in which the efficiency of the GC method is improved without any loss in the disparity estimation performance by dividing all the disparity levels within the stereo image hierarchically into a series of bilateral disparity structures of increasing fineness. To address the well-known foreground fattening effect, a disparity refinement process is proposed comprising a fattening foreground region detection procedure followed by a disparity recovery process. The efficiency and accuracy of the HBDS-based GC algorithm are compared with those of the conventional GC method using benchmark stereo images selected from the Middlebury dataset. In addition, the general applicability of the proposed approach is demonstrated using several real-world stereo images.

Current graph embedding frameworks of supervised dimensionality reduction often preserve the intraclass local structures and maximize the interclass variance. However, this strategy fails to provide adequate results when strict within-class multimodalities contradict between-class separations. In this paper, we propose Hypersphere Distribution Discriminant Analysis (HDDA), which determines the affinity by considering not only within-class local structure but also the heteropoint distribution in the neighborhood space. If the heteropoint distribution is relatively high in the feature space, this pair should be mapped apart to avoid mixing problems. By taking both the distribution of heteropoints and the distance into account, HDDA shows more effective results compared to the state-of-the-art methods.

This paper presents an inertial-sensor-based wearable device and its associated stride detection algorithm to analyze gait information for patients with Alzheimer's disease (AD). The wearable gait analysis device is composed of a triaxial accelerometer, a microcontroller, and an RF wireless transmission module. To validate the effectiveness of the proposed device and algorithm, nine AD patients and three healthy controls were recruited to participate a gait analysis experiment. They were asked to mount the device on their foot and walk along a straight line of 40 meters at normal speed. The stride detection algorithm, consisting of procedures of data collection, signal preprocessing, and stride detection, has been developed for acquiring gait feature information from acceleration signals. The advantages of this wearable gait analysis device include the following: 1) It can be used anywhere without any external device, and 2) the stride detection algorithm can acquire gait feature information from acceleration signals automatically and effectively. Experimental results show that the AD patients exhibited a significantly shorter mean stride length and slower mean gait speed than those of the healthy controls. No significant differences in mean stride frequency and mean cadence were observed in the two groups. The variability in the percentage of the stance phase of the AD patients was slightly greater than that of the healthy controls. Based on the above results and discussions with physicians, we conclude that the proposed wearable gait analysis device is a promising tool for automatically analyzing gait information which can serve as indicators for early diagnosis of AD.

In order to continuously keep sight of environments, taking long-time surveillance videos is required. Thus, enormous storage of videos and video browsing become important problems. To solve these problems, we propose an online surveillance video synopsis method to rearrange positions and appearing frames of foreground objects chronologically in real-time. Comparing with traditional synopsis approaches, our method can directly be applied to streaming surveillance videos and generate synopsis videos without pre-screening entire videos during the synopsis processing.

Human respiration has been reported as having great relevance to the emotional state. However, the respiration signal obtained using traditional biosensors not only reflects changing emotions and emotion intensities, but also contains motion artifacts. The resulting information ambiguity limits the practical application of the respiration signal as a means of reliable affective state appraisal. Thus, the present study proposes a method for extracting representative Emotion Elicited Segments (EESs) from the respiration signal such that the affective state of the individual can be more reliably determined. The EES extraction process involves the combination of the following procedures, namely, 1) Mutual Information-Based Emotion Relevance Feature Ranking based on the Dynamic Time Warping Distance (MIDTW), and 2) Constraint-based Elicited Segment Density (CESD) analysis. Due to the innate regularity of the respiration signal, the respiration pattern under a particular emotional state would be relatively quasihomogeneous over time. Accordingly, a parameter-free Respiration quasi-Homogeneity Segmentation (RHS) algorithm is proposed for partitioning the respiration signal into quasi-homogenous segments from which the EESs can then be extracted. The experimental results obtained for five prototypical emotions (i.e., “love,” “sadness,” “joy,” “anger,” and “fear”) show that the proposed segmentation/extraction methodology enables the EESs to be reliably identified and yields an average classification rate of 88 percent.

This paper explores the detection of emotional levels, high, medium and low from ECG signals. Features of ECG are extracted from frequency domain, time domain and nonlinear method and normalized by z-score method. Then SFS feature selection is applied followed by LDA feature transformation. After that, the transformed features are applied to KNNR classifier. According to our results, it was found that subjects of different characteristics reveal different biosignal responses. While subjects are regarded as optimistic characteristics, they have higher responses on positive films. On the other hand, the pessimistic subjects have higher responses on negative films. When classifiers are established separately for optimistic and pessimistic subjects, we can achieve the recognition rate of 97.8%, where 7 features are selected, and 94.0%, where 8 features are selected, for optimistic and pessimistic groups respectively. When the classification is built from all the subjects, the recognition rate is reduced slightly, but it can still maintain a recognition rate of 90.4%.

In this paper, we propose a new scheme which combines two algorithms to detect epileptic seizure in the grouped multi-channel EEG signals. For the proposed scheme, a recent technique, Independent Component Analysis (ICA), is first adapted to separate blind sources and extract feature from grouped EEG signals. Then, Wavelet transform is followed for multi resolution and multi-level analysis on those primary signals extracted by ICA. Finally, a threshold method based on wavelet transform again is applied to detect the epileptic seizure. A series of experiments using different method combination are conducted and the experimental results show that the proposed method has a superior quality.

One of the major goals of healthcare systems is to automatically monitor patients of special needs and alarm the caregivers for providing assistant. In this paper, an efficient single-camera multidirectional wheelchair detector based on a cascaded decision tree (CDT) is proposed to detect a wheelchair and its moving direction simultaneously from video frames for a healthcare system. Our approach combines a decision tree structure and boosted-cascade classifiers to construct a new CDT that can perform early confidence decisions in a hierarchical manner to rapidly reject nonwheelchairs and decide the moving directions. We also impose the tracking history to guide detection routes in the CDT to further reduce detection time and increase detection accuracy. The experiments show over 92% detection rate under cluttered scenes.

Due to the rapid growth of the elderly population, improving specific aspects of elderly healthcare has become more important. Sleeping care systems for the elderly are rare. In this paper, we propose a visual context-aware-based sleeping-respiration measurement system that measures the respiration information of elderly sleepers. Accurate respiration measurement requires considering all possible contexts for the sleeping person. The proposed system consists of a body-motion-context-detection subsystem, a respiration-context-detection subsystem, and a fast motion-vector-estimation-based respiration measurement subsystem. The system yielded accurate respiratory measurements for our study population.

This paper presents an interaction-embedded hidden Markov model (IE-HMM) framework for automatically detecting and classifying individual human behaviors and group interactions. The proposed framework comprises a switch control (SC) module, an individual duration HMM (IDHMM) module, and an interaction-coupled duration HMM (ICDHMM) module. By analyzing the relative distances between the various participants in each scene, and monitoring the duration for which these distances are maintained, the SC module assigns each participant to an individual behavior unit (comprising a single participant) or an interaction behavior unit (comprising two or more participants). The individual behavior units are passed to the IDHMM module, which classifies the corresponding human behavior in accordance with the pose, motion, and duration information using duration HMM (DHMM). Similarly, the interaction behavior units are dispatched to the ICDHMM module, where the corresponding interaction mode is classified using an integrated scheme comprising multiple coupled-duration HMM (CDHMM), in which each state has an embedded coupled HMM (CHMM). The validity of the IE-HMM framework is confirmed by analyzing the human actions and interactions observed in a nursing home environment. The results confirm that the atomic behavior unit concept embedded in the SC module enables the IE-HMM framework to recognize multiple concurrent actions and interactions within a single scene. Overall, it is shown that the proposed framework has a recognition performance of 100% when applied to the analysis of individual human actions and 95% when applied to that of group interactions.

An emotion recognition system with consideration of facial expression and physiological signals is proposed in this paper. A specific designed mood induction experiment is performed to collect facial expressing images and physiological signals of subjects. We detected 14 feature points and extracted 12 facial features from facial expression images. Meanwhile, we measure the skin conductivity, finger temperature and heart rate from the subject. Both facial and physiological features are adopted to train the classifiers. Two learning vector quantization (LVQ) neural networks were applied to classify four emotions: love, joy, surprise and fear. Experimental results show the proposed recognition system is able to identify four emotions by facial expressions, physiological signals, and both of them.

Tracking the patella movement trajectory during the bending process of the knee is one essential step to knee pain diagnosis. In order for tracking patella, correct segmentation of the femur and patella from the axial knee MR image is indispensable. But the strong adhesion of the soft tissue around femur and patella, the gray-level similarities of adjacent organs, and the non-uniform gray intensity due to the degradation of the magnetic propagation make the MR image segmentation challenging. In this paper, we proposed a mechanism combining contextual constraint neural network (CCNN) and level set evolution to segment the femur and patella. The segmentation can be divided into two phases. In the first phase SOM and CCNN are applied to extract initial contours of the femur and patella. Consequently in the second phase, modified level set evolution is performed, with the extracted contours as the initial zero level set contour, to accomplish the segmentation of the femur and patella. Our experimental results show that the femur and patella can be correctly segmented for tracking patella movement.

This paper presents a computer-aided prescreening and storage system, which automatically prescreens the mass regions from mammograms and based on the results, performs a progressive compression in the storage. This is performed in two subsystems called mass screening subsystem and mass feature reserved compression subsystem. In the first subsystem, breast region is firstly extracted from images, followed by Gradient Enhancement and Median Filtering. Then, 19 texture features are calculated from 32*32 pixel blocks on the extracted breast region, and suboptimal feature subset is extracted. Then SVM classifier is employed for classifying the regions into mass, breast without masses and background. In the second subsystem, Vector Quantization GHNN (Grey-based Competitive Hopfield neural network) is applied on the three regions with different compression rates according their importance factors so as to reserve important features and simultaneously reduce the size of mammograms for storage efficiency.

This paper develops the fusion of audio and video features by Dempster-Shafer theory for seizure detection. In audio analysis, Mel Frequency Cepstral Coefficient (MFCC) and Zero-Crossing Rate (ZCR) are applied to Hidden Markov Model (HMM) for audio type classification and probability computation. The results are transferred to belief of evidence and combined with the results from videos. Results have been tested by data obtained from several seizure patients and showed promising results.

This study presents a computer-aided diagnosis system using sequential forward floating selection (SFFS) with support vector machine (SVM) to diagnose gastric histology of Helicobacter pylori (H. pylori) from endoscopic images. To achieve this goal, candidate image features associated with clinical symptoms are extracted from endoscopic images. With these candidate features, the SFFS method is applied to select feature subsets, which perform the best classification results under SVM with respect to different histological features. By using the classifiers obtained from the feature subsets, a new diagnosis system is implemented to provide physicians with H. pylori -related histological results from endoscopic images.

As people live longer in the new millennium, it becomes a necessity to develop affordable technologies to improve the quality of life. This paper provides an overview of such techniques in drug delivery and clinical care, namely: (i) health care from organ functions to the cell behaviors; (ii) the development of implantable biosensors and instrumentation; (iii) the process and visualization systems; and (iv) nanotechnology for drug delivery. The goal is to stimulate cross-disciplinary research in the circuits and systems society aimed towards optimal clinical care and personalized therapeutic intervention.

In this paper, a liver disease diagnosis based on Gabor filters is proposed. Three kinds of liver diseases are identified: cyst, hepatoma and cavernous hemangioma. The diagnosis scheme includes two steps: features extraction and classification. The features derived from Gabor filters are obtained from the ROIs among the normal and abnormal CT images. In the classification step the SVM classifier is used to discriminate the different liver disease types. Finally the receiver operating characteristic curve is employed to evaluate the performance of the diagnosis system. The effectiveness of the proposed method is demonstrated through experimental results on CT images including 76 liver cysts, 30 hepatomas, and 40 cavernous hemangiomas. From the results, we can observe that the discrimination rate of cyst is higher than the other diseases, and the classification accuracy decreases slightly between cavernous hemangiomas and hepatomas. However, a normal region can be discriminated from all of these diseases entirely.

This paper presents an intelligent guiding bulletin board system (iGBBS), which is based on vision-interactive and multiple key word-spotting technology. The system is aimed to provide different kinds of multimedia human-computer interaction (MMHCI) for users under different requirements. At first, a real-time front-view face detection using Harr-like features is used to decide when iGBBS should wake up and become interactive with the user. After system initialization, some feature points within the detected face area are going to be found. Then the orientation of user's head will be estimated via pyramidal Lucas-Kanade optical flow tracking. In addition, spotting the keyword from user's utterance with some related augmented reality responses would be provided as well. The performance of vision-interaction in iGBBS could be reached to 20 fps under Pentium IV 1G Hz PC. The error rate of multiple key word-spotting interaction in iGBBS is about 36.2% and people can get the right response in 2.76 times search averagely. With the comparison to the traditional guiding system, bulletin board, or other non-vision-based input devices system, such like gloves or markers, our system offers a simple, useful and economical solution for the realtime interaction between the user and computer

This paper presents a new adult image identification system. We propose the online skin tone sampling mechanism based on face detection. The object detection based on machine learning approach is used to locate the face region in the target image. According to the face color information, the skin chromatic distribution is extracted to detect the skin objects in the target image. The texture feature, coarseness, is also utilized to acquire the more accurate skin regions. Then, the back propagation neural network is used to integrate several low-level and reliable geometrical constraints into a classifier to inspect the skin regions further. Finally, the mug shot exclusion procedure is applied to promote the system performance. The experimental results show our method is satisfactory for adult image detection.

This paper presents a real-time vision-interactive guiding system, which could be interactive with users based on the computer vision technology. A real-time front-view face detection using Harr-like features is used to decide when the guiding system should wake up and become interactive with the user. After system initialization, some feature points within the detected face area are going to be found. Then the orientation of user's head will be estimated via pyramidal Lucas-Kanade optical flow tracking. Compared to the traditional guiding system, our system has more flexibility. Guiding providers could choose a suitable display equipment according to the surround environment. With the comparison to other non-vision-based input devices system, such like gloves or markers, our system offers a simple, useful and economical solution for the real-time interaction between the user and computer.

This paper presents an RFID based psychiatric patient tracking system in a psychiatric patient care center. In this system, RFID field generators and readers are installed in the environment, near the entrance and exit points for tracking the patients. Every patient is required to wear the tag. When a tag receives the call from a field generator, it responds to a reader with the ID of the field generator issuing the calling. Based on the field generator IDs a tag responds, the location of the patient associated with the tag can be approximately estimated. By so doing, not only the patient's moving patterns can be recorded but also the event of the patient leaving the main gate of the care center be detected. Besides the localization issue, in this paper we also employed a graphic coloring approach for solving the interference caused from field generator located one near another. Experimental results show that this algorithm has significantly improved the reliability of the tracking system

With the progress of advanced technology, people anticipate better quality in the medical health care service. Due to more robust requirements in medical environments, higher availability and reliability are expected. This paper describes a recovery scheme for disconnection failures in a Web-based medical teleconsultation system. With the support, the disconnected clients can re-enter the system to resume the unfinished discussion using the synchronization mechanisms. The experiments have shown that the mechanism performed successfully. The lazy synchronization further reduced the required time that the clients had to wait for joining the consultation after network disconnections

Presently, radiologists used to rely on contrast-injection to acquire the contrast-enhanced breast magnetic resonance imaging (MRI), in order to improve the accuracy of breast cancer screening. Instead of contrast-injection, this paper proposed a spectral signature detection technology, constrained energy minimization (CEM), which could successfully classify breast MRIs into four major tissues (fatty tissue, glandular tissue, tumor and muscle) and show the classified results in high contrast images. After compared with a specific subspace projection operator called orthogonal subspace projection (OSP), the commonly used C-means (CM) algorithm and real contrast-injected breast MRIs, the results show that the high contrast images generated by CEM have superior quality.

In this paper, we propose a new region-of-interest (ROI) coding method called fractal based JPEG2000 ROI coding for the further improvement of JPEG2000 used in mammograms. This method provides the advantage that it is not necessary for user to provide any ROI information before coding.

This paper proposes an uneven block-sized information included marker (UBIIM) in the use of JPEG 2000 to achieve both compression efficiency and error resilience capability. Included in the UBIIM approach is the marker adopted in JPEG 2000 devised to incorporate also error detection and correction capabilities. The UBIIM consists of VS Index and TR Index, where VS Index is used to indicate a value support block, one of the neighboring blocks of the erroneous block of which the pixel values will be most similar as the erroneous block. TR Index is used to indicate a texture reference block, the block in the corresponding position of the erroneous block in the same subband of one-higher level of which the pixel value distribution is similar as the erroneous block. When errors occur, a new block is formed to replace the erroneous block by value support block and texture reference block, to achieve error concealment. The sizes of the blocks in UBIIM are determined based on the importance factor of various levels of subbands, that is, different levels have different block sizes. With such approach, the possible overhead due to the addition of markers between every two blocks could be dramatically reduced, while error resilience capability significantly improved.

Motion vector plays one significant feature in MPEG-4 video object segmentation. However, the motion vector in this application is required to represent the actual motion displacement, rather than regions of visually significant similarity. Optical flow back-projection (OFB), which back-projects optical flows in a cluster to restore the region's motion vector from gradient-based optical flows, is proposed to obtain genuine motion displacement. The back-projection is performed based on minimizing the projection mean square errors of the motion vector on gradient directions. As optical flows of various magnitudes and directions provide various degrees of reliability in the genuine motion restoration, the optical flows to be used in the OFB are selected based on high gradient. With this approach, the interference by similar blocks, which frequently results in irrational motion displacement in the traditional block matching approach, can be avoided and more reliable motion vectors can be obtained.

Identifying abdominal organs is one of the essential steps in visualizing organ structure to assist in teaching, clinical training, diagnosis, and medical image retrieval. However, due to partial volume effects, gray-level similarities of adjacent organs, contrast media affect, and the relatively high variations of organ position and shape, automatically identifying abdominal organs has always been a high challenging task. To conquer these difficulties, this paper proposes combining a multimodule contextual neural network and spatial fuzzy rules and fuzzy descriptors for automatically identifying abdominal organs from a series of CT image slices. The multimodule contextual neural network segments each image slice through a divide-and-conquer concept, embedded within multiple neural network modules, where the results obtained from each module are forwarded to other modules for integration, in which contextual constraints are enforced. With this approach, the difficulties arising from partial volume effects, gray-level similarities of adjacent organs, and contrast media affect can be reduced to the extreme. To address the issue of high variations in organ position and shape, spatial fuzzy rules and fuzzy descriptors are adopted, along with a contour modification scheme implementing consecutive organ region overlap constraints. This approach has been tested on 40 sets of abdominal CT images, where each set consists of about 40 image slices. We have found that 99% of the organ regions in the test images are correctly identified as its belonging organs, implying the high promise of the proposed method.

This paper presents a new spectral signature detection approach to magnetic resonance (MR) image classification. It is called constrained energy minimization (CEM) method, which is derived from the minimum variance distortionless response in passive sensor array processing. It considers a bank of spectral channels as an array of sensors where each spectral channel represents a sensor and object spectral signature in multispectral MR images are viewed as signals impinging upon the array. The strength of the CEM lies on its ability in detection of spectral signatures of interest without knowing image background. The detected spectral signatures are then used for classification. The CEM makes use of a finite impulse response (FIR) filter to linearly constrain a desired object while minimizing interfering effects caused by other unknown signal sources. Unlike most spatial-based classification techniques, the proposed CEM takes advantage of spectral characteristics to achieve object detection and classification. A series of experiments is conducted and compared with the commonly used c-means method for performance evaluation. The results show that the CEM method is a promising and effective spectral technique for MR image classification.

Compared to object-based registration, feature-based registration is much less complex. However, in order for feature-based registration to work, the two image stacks under consideration must have the same acquisition tilt angle and the same anatomical location - two requirements that are not always fulfilled. In this paper, we propose a technique that reconstructs two sets of medical images acquired with different acquisition angles and anatomical cross sections into one set of images of identical scanning orientation and positions. The space correlation information among the two image stacks is first extracted and is used to correct the tilt angle and anatomical position differences in the image stacks. Satisfactory reconstruction results were presented to prove our points.

In this paper, we propose an eigen-based fuzzy C-means (FCM) method for color object segmentation. After sampling a few color samples, we can form the sampled covariance matrix and its related eigenvectors of the desired color space. Then, we transform the original color space into signal and noise planes of the desired color. Followed the transformation, the proposed eigen-based FCM algorithm is finally applied to the signal and noise subspaces individually. After few iterated classification processes, the desired color objects can be easily identified without using any threshold procedure. Inspecting the segmented results, the desired color objects without any pre- and post-processes can be extracted easily and robustly.

Function approximation has been found in many applications. The radial basis function (RBF) network is one approach which has shown a great promise in this sort of problems because of its faster learning capacity. A traditional RBF network takes Gaussian functions as its basis functions and adopts the least-squares criterion as the objective function, However, it still suffers from two major problems. First, it is difficult to use Gaussian functions to approximate constant values. If a function has nearly constant values in some intervals, the RBF network will be found inefficient in approximating these values. Second, when the training patterns incur a large error, the network will interpolate these training patterns incorrectly. In order to cope with these problems, an RBF network is proposed in this paper which is based on sequences of sigmoidal functions and a robust objective function. The former replaces the Gaussian functions as the basis function of the network so that constant-valued functions can be approximated accurately by an RBF network, while the latter is used to restrain the influence of large errors. Compared with traditional RBF networks, the proposed network demonstrates the following advantages: (1) better capability of approximation to underlying functions; (2) faster learning speed; (3) better size of network; (4) high robustness to outliers.

An efficient texture representation for unsupervised segmentation is addressed based on the concept of Wold decomposition. Textures are described by the wavelet tuned to various scales and rotations to describe its deterministic component, and by the autoregressive model to describe its indeterministic component. The wavelet features and the AR parameters capturing the perceptual properties, "periodicity", "directionality", and "randomness", respectively, have been proved to be consistent with human texture perception. The performance of our approach is demonstrated on Brodatz textures and natural textured images.

In this correspondence, a method is proposed for estimating the surface orientation of a planar texture under perspective projection based on the ridge of a two-dimensional (2-D) continuous wavelet transform (CWT). We show that an analytical solution of the surface orientation can be derived from the scales of the ridge surface. A comparative study with an existing method is given.

A common assumption of the shape from texture problem is that a perceived image mainly contains only one type of texture with the same surface orientation. Unfortunately, a natural image is often composed of more than one textures. In order to solve the shape from texture problem in a practical manner, we need to segment 3D textured images. In this paper, we propose a new algorithm for the task. We estimate the local surface orientations from the the scales of the ridge points of continuous wavelet transform. Then, the local surface orientations are used as the features for texture segmentation. Textured images synthesized from Brodatz's album and several natural images demonstrate the performance of our method.

We propose a new shape from texture method based on the ridge of continuous wavelet transform. This method determines the orientations of a planar surface in a direct way under the perspective projection model. The variations of the image projected from a planar surface can be accurately characterized by the ridge of the continuous wavelet transform. The ridge of the 1-D signal and 2-D image are represented as a ridge curve and ridge plane, respectively. Ridges represent the energy concentration in the time-frequency plane where the energy is a local maxima. We show that the ridge of the projected image is a parabolic plane with a rotation angle equal to the tilt angle of the planar surface. The ridge is then rotated with the angle such that the slant effect appears in the X-axis and plays no role along the Y-axis. As a result, the rotated ridge plane can be regarded as the plane composed of many 1-D ridge curves. The slant angle of the 2-D image is thus obtained from the derived slant angle of the 1-D signal. A voting method and a curve fitting method are developed to obtain the slant angle of the 1-D signal. Several synthetic and real-world images have demonstrated the robustness and accuracy of our method.

The performance capability of quadratic Hebbian type associative memories (QHAM's) in the presence of interconnection faults is examined, and equations for predicting the probability of direct convergence P(dc) given a fraction of interconnection faults are developed. The interconnection faults considered are the equivalent of open circuit and short circuit synaptic interconnections in electronic implementations. Our results show that a network with open circuit interconnection faults has a higher probability of direct convergence P (dc) than a network with short circuit interconnection faults, when the fraction of failed interconnections p is small and the short circuit signal G is large. Certain values of G are found to have only mild effects on network performance degradation. Network reliability characteristics taking the generalization capability into account are also analyzed. All of these results are compared with those of Hebbian type associative memories (HAM's), which have linear association network models. Our results indicate that QHAM's have much higher network capacity and fault tolerance capability in the presence of interconnection faults. However, the fault tolerance to input errors in QHAM's is much less than that of HAM's.

The performance of Hebbian-type associative memories (HAMs) in the presence of faulty (open- and short-circuit) synaptic interconnections is examined and equations for predicting network reliability are developed. The results show that a network with open-circuit interconnection faults has a higher probability of direct convergence than a network with short-circuit interconnection faults when the fraction of failed interconnections is small and the short-circuit signal is large. The results are extended to the case where network attraction radius is considered. Under certain assumptions, it is found that the expected numbers of neurons with b, b-1, b-2,. . .,1 input error bits in their state update are equal. Because of the capability of error correction, an asynchronous HAM is also found to have a higher probability of direct convergence than a synchronous HAM. Using these results, network reliability and generalization capability can be estimated when both the interconnection faults and the number of error bits in the probe vectors are taken into account.

The performance of Hebbian-type associative memories (HAMs) in the presence of faulty interconnections is examined, and equations for predicting network reliability are developed. Several HAM operating modes, i.e. zero or nonzero autoconnections (ZA or NZA), bipolar or unipolar network inputs, asynchronous or synchronous operation, have been investigated. Networks with NZA and bipolar inputs have the best performance in both capacity and reliability. An equation relating the probability of direct one-step convergence,Pdc, to the percentage of failed connections is derived. For a connection failure rate of up to 50%, this equation can estimatePdcaccurately. The reliability becomes poorer when the network size increases; hence, reducing the percentage of failed connections becomes very important in physical implementations of large networks. Also, as the number of stored vectors increases, the network reliability decreases. Asynchronous-mode and synchronous-mode operation have the samePdcand reliability when there are no error bits in the probe vector, but networks operating in the asynchronous mode have higher Pdcand better reliability when the number of input errors increases. Reliability measures for second-order HAMs are also investigated