Automatic whole-brain extraction from magnetic resonance images (MRI), also known as skull stripping, is a key component in most neuroimage pipelines. As the first element in the chain, its robustness is critical for the overall performance of the system. Many skull stripping methods have been proposed, but the problem is not considered to be completely solved yet. Many systems in the literature have good performance on certain datasets (mostly the datasets they were trained/tuned on), but fail to produce satisfactory results when the acquisition conditions or study populations are different. In this paper we introduce a robust, learning-based brain extraction system (ROBEX). The method combines a discriminative and a generative model to achieve the final result. The discriminative model is a Random Forest classifier trained to detect the brain boundary; the generative model is a point distribution model that ensures that the result is plausible. When a new image is presented to the system, the generative model is explored to find the contour with highest likelihood according to the discriminative model. Because the target shape is in general not perfectly represented by the generative model, the contour is refined using graph cuts to obtain the final segmentation. Both models were trained using 92 scans from a proprietary dataset but they achieve a high degree of robustness on a variety of other datasets. ROBEX was compared with six other popular, publicly available methods (BET, BSE, FreeSurfer, AFNI, BridgeBurner, and GCUT) on three publicly available datasets (IBSR, LPBA40, and OASIS, 137 scans in total) that include a wide range of acquisition hardware and a highly variable population (different age groups, healthy/diseased). The results show that ROBEX provides significantly improved performance measures for almost every method/dataset combination.

From May 1985 through July 1990, 28 episodes of Vibrio vulnificus infection in 27 patients were encountered in five major hospitals in Taiwan. The ages of patients ranged from 19 to 76 years; the ratio of male to female patients was 2:1. Eighteen episodes manifested as bacteremia and eight as wound infections alone. One patient each developed gastroenteritis and pneumonia after nearly drowning. Twenty-three patients exhibited skin manifestations. Twenty patients had underlying diseases. All patients were treated with antibiotics, and 14 also underwent some form of surgical treatment (incision and drainage, fasciotomy, debridement, or amputation). Thirteen of the 28 episodes were preceded by precipitating factors; most were due to ingestion of seafood or exposure of abraded skin to salt water. Ten of the 18 septicemic patients died--most within 48 hours of hospitalization. One patient without bacteremia who had a wound infection died. Results of in vitro susceptibility studies suggested that ampicillin or a third-generation cephalosporin would be effective. Susceptibility to aminoglycosides was observed for greater than 90% of isolates. We recommend combined therapy with a third-generation cephalosporin or ampicillin and an aminoglycoside along with appropriate surgical therapy for the treatment of V. vulnificus infection.

While some trends in antimicrobial resistance rates are universal, others appear to be unique for specific regions. In Taiwan, the strikingly high prevalence of resistance to macrolides and streptogramin in clinical isolates of gram-positive bacteria correlates with the widespread use of these agents in the medical and farming communities, respectively. The relatively low rate of enterococci that are resistant to glycopeptide does not parallel the high use of glycopeptides and extended-spectrum beta-lactams in hospitals. The evolving problem of extended-spectrum beta-lactamase-producing Escherichia coli and Klebsiella pneumoniae isolates is substantial, and some unique enzymes have been found. Recently, some gram-negative bacteria (e.g., Pseudomonas aeruginosa and Acinetobacter baumannii) that are resistant to all available antimicrobial agents including carbapenems have emerged.

The clinical manifestations of SARS showed high variability, and were related to the underlying health status of individual patients. Importantly, the development of ARDS was associated with significant mortality, despite aggressive therapy.

Seroepidemiology of Klebsiella pneumoniae was determined for 1000 nonrepetitive K. pneumoniae isolates collected by a medical center in Taiwan during 1993-1997. Of these, 630 isolates (63%) were from community-acquired infections; the rest were from hospital-acquired infections. The isolates were serotyped according to capsular antigen by countercurrent immunoelectrophoresis. About 77% were typeable. Serotypes K1 and K2 accounted for 21.7% and 9.3% of the isolates, respectively, followed by K57 (5.1%), K54 (4.2%), K21 (3. 3%), and K16 (3%). The frequency of serotype K1 among bacteremic isolates (30.8%) far exceeded that reported by other investigators worldwide. Molecular typing of random K1 isolates by pulsed-field gel electrophoresis revealed several different pulsotypes, suggesting a nonclonal spread. This study indicates that a Klebsiella vaccine developed in Europe is not optimal for use in Taiwan because it does not contain the most predominant serotypes-K1, K54, and K57.

No abstract

No abstract

Cognitive performance relies on the coordination of large-scale networks of brain regions that are not only temporally correlated during different tasks, but also networks that show highly correlated spontaneous activity during a task-free state. Both task-related and task-free network activity has been associated with individual differences in cognitive performance. Therefore, we aimed to examine the influence of cognitive expertise on four networks associated with cognitive task performance: the default mode network (DMN) and three other cognitive networks (central-executive network, dorsal attention network, and salience network). During fMRI scanning, fifteen grandmaster and master level Chinese chess players (GM/M) and fifteen novice players carried out a Chinese chess task and a task-free resting state. Modulations of network activity during task were assessed, as well as resting-state functional connectivity of those networks. Relative to novices, GM/Ms showed a broader task-induced deactivation of DMN in the chess problem-solving task, and intrinsic functional connectivity of DMN was increased with a connectivity pattern associated with the caudate nucleus in GM/Ms. The three other cognitive networks did not exhibit any difference in task-evoked activation or intrinsic functional connectivity between the two groups. These findings demonstrate the effect of long-term learning and practice in cognitive expertise on large-scale brain networks, suggesting the important role of DMN deactivation in expert performance and enhanced functional integration of spontaneous activity within widely distributed DMN-caudate circuitry, which might better support high-level cognitive control of behavior.