In the field of MRI super-resolution, training an image upscaling network under a pixel-oriented cost function (e.g., Mean-Intensity-Error) has proven to boost the signal-to-noise ratio. However, these types of cost functions tend to miss high-frequency details and fail to achieve an ideal sharpness, which is a pivotal image property for clinical applications to make diagnoses. To address this issue, the cost function of these upscaling networks typically includes a perceptual loss function, which is well recognized for the reconstruction of textures and enhancing sharpness, in addition to a pixel-oriented one. In this paper, we investigate the effect of perceptual loss on several MRI super-resolution metrics. We train UNet architecture under two loss function scenarios: One only including a pixel-oriented loss function, and the other a fusion of pixel-oriented and perceptual losses. We then employ an ablation study using a mixed effect model on a comprehensive set of evaluation criteria to measure the significance of change upon the inclusion of perceptual loss. Our results show that even though perceptual loss substantially shifts the networks towards outputting sharper images, it only causes negligible performance degradation in the accuracy of the reconstructed regions of interest, which can be alleviated using proper hyperparameter tuning.

This work presents an interventional planning software to be used in conjunction with a robotic manipulator to perform transrectal MR guided prostate biopsies. The interventional software was designed taking in consideration a generic manipulator used under the two modes of operation: side-firing and end-firing of the biopsy needle. Studies were conducted with urologists using the software to plan virtual biopsies. The results show features of software relevant for operating efficiently under the two modes of operation.

Prostate biopsy is considered as a definitive way for diagnosing prostate malignancies. Urologists are currently moving towards MR-guided prostate biopsies over conventional transrectal ultrasound-guided biopsies for prostate cancer detection. Recently, robotic systems have started to emerge as an assistance tool for urologists to perform MR-guided prostate biopsies. However, these robotic assistance systems are designed for a specific clinical environment and cannot be adapted to modifications or changes applied to the clinical setting and/or workflow. This work presents the preliminary design of a cable-driven manipulator developed to be used in both MR scanners and MR-ultrasound fusion systems. The proposed manipulator design and functionality are evaluated on a simulated virtual environment. The simulation is created on an in-house developed interventional planning software to evaluate the ergonomics and usability. The results show that urologists can benefit from the proposed design of the manipulator and planning software to accurately perform biopsies of targeted areas in the prostate.

The emerging potential of augmented reality (AR) to improve 3D medical image visualization for diagnosis, by immersing the user into 3D morphology is further enhanced with the advent of wireless head-mounted displays (HMD). Such information-immersive capabilities may also enhance planning and visualization of interventional procedures. To this end, we introduce a computational platform to generate an augmented reality holographic scene that fuses pre-operative magnetic resonance imaging (MRI) sets, segmented anatomical structures, and an actuated model of an interventional robot for performing MRI-guided and robot-assisted interventions. The interface enables the operator to manipulate the presented images and rendered structures using voice and gestures, as well as to robot control. The software uses forbidden-region virtual fixtures that alerts the operator of collisions with vital structures. The platform was tested with a HoloLens HMD in silico. To address the limited computational power of the HMD, we deployed the platform on a desktop PC with two-way communication to the HMD. Operation studies demonstrated the functionality and underscored the importance of interface customization to fit a particular operator and/or procedure, as well as the need for on-site studies to assess its merit in the clinical realm.

Clouds are becoming prevalent service providers because of their low upfront costs, rapid application deployment, and high scalability. Many users outsource their sensitive data and services to cloud providers. Users frequently access these sensitive services through devices and connections that are vulnerable to thieving and eavesdropping. Therefore, users are desperate of robust security measures to protect their data and services privacy in clouds. In particular, robust authentication techniques are demanded by users for safe access to cloud services. One technique is to utilize multiple authentication factors (a.k. A multi-factor authentication) to access cloud services. However, the challenge is that the multi-factor authentication technique is not effective as it causes user frustration and fatigue. To address this challenge, in this study, we propose a multi-factor authentication architecture that aims at minimizing the perceived authentication hardship for cloud users while improving the security of the authentication. To achieve the goal, our authentication architecture suggests a progressive manner to leverage access to different levels of cloud services. At each level, the architecture asks for authentication factors by considering the perceived hardship for users. To increase the security and user convenience, the architecture also considers implicit authentication factors in addition to the explicit factors. Our evaluation results indicate that authentication using the proposed architecture decreases the users' perceived hardship up to 29% in compare with other methods. The results also reveal that our proposed architecture adapts the authentication difficulty based on the user condition.

The unprecedented use of social networking, the Web and other related technologies by all fringes of population, has not been accompanied by an adequate awareness of its potential for abuse, privacy violation and cybercrimes. Furthermore, technology nowadays provides extensive capabilities which often times clash with society's laws and ethics. In particular, modern societies have certain expectations of privacy; that may be contradicted by the capabilities of new technology. It seems then incumbent upon computer science instructors to raise awareness of these risks and threats among the students they teach. The purpose of this session is to provide a forum for discussing technologies that are widely used by CS and non CS professionals alike, and that may well be at variance with society's privacy expectations. A second purpose of this session is to introduce instructional modules currently under development that may assist instructors in addressing some of the relevant themes in courses, in particular privacy in social networks, on the internet, in mobile apps and also privacy in electronic voting and law enforcement. We expect that attendees will provide varying opinions and insight that will help shape the course offering. In addition, attendees will be invited to share their experience teaching or talking about privacy issues and the approaches, resources and techniques they have used in presenting the topic to students.

Cardiac gating or breath-hold MRI acquisition is challenging. In particular, data collected in a short amount of time might be insufficient for the diagnosis of patients with impaired breath-holding capabilities and/or arrhythmia. A major challenge in cardiac MRI is the motion of the heart itself, the pulsate blood flow, and the respiratory motion. Furthermore, the motion of the diaphragm in the chest moving up and down gets translated to the heart when a patient breathes. Therefore, artifacts arise due to the changes in signal intensity or phase as a function of time, resulting in blurry images. This paper describes a novel reconstruction strategy for real time cardiac MRI without requiring the use of an electro-cardiogram or of breath holding. In this research we focused on automation and evaluation of the performance of our proposed method in real time MRI data to ensure a good basis for the signal extraction. Hence, it assists in the reconstruction. The proposed method enables one to extract cardiac beating waveforms directly from real-time cardiac MRI series collected from freely breathing patients and without cardiac gating. Our method only requires minimal user involvement as initialization step. Thereafter, the method follows the registered area in every frame and updates itself.

Face recognition is a challenging research topic, especially when the training (gallery) and recognition (probe) images are acquired using different cameras under varying conditions. Even a small noise or occlusion in the images can compromise the accuracy of recognition. Lately, sparse encoding based classification algorithms gave promising results for such uncontrollable scenarios. In this paper, we introduce a novel methodology by modeling the sparse encoding with weighted patches to increase the robustness of face recognition even further. In the training phase, we define a mask (i.e., weight matrix) using a sparse representation selecting the facial regions, and in the recognition phase, we perform comparison on selected facial regions. The algorithm was evaluated both quantitatively and qualitatively using two comprehensive surveillance facial image databases, i.e., SCfaceandMFPV, with the results clearly superior to common state-of-the-art methodologies in different scenarios. Publisher’s Note: This paper, originally published on 24 December 2013, was replaced with a revised version on 11 June 2014. If you downloaded the original PDF but are unable to access the revision, please contact SPIE Digital Library Customer Service for assistance.

In this paper, we propose a new algorithm to estimate a super-resolution image from a given low-resolution image, by adding high-frequency information that is extracted from natural high-resolution images in the training dataset. The selection of the high-frequency information from the training dataset is accomplished in two steps: a nearest-neighbor search algorithm is used to select the closest images from the training dataset, which can be implemented in the GPU, and a sparse-representation algorithm is used to estimate a weight parameter to combine the high-frequency information of selected images. This simple but very powerful super-resolution algorithm can produce state-of-the-art results. Qualitatively and quantitatively, we demonstrate that the proposed algorithm outperforms existing common practices.

This paper describes complementary software- and hardware-based approaches for handling overlapping register lifetimes that occur in modulo scheduled loops. Modulo scheduling takes the N-instructions in a loop body and constructs an M-stage software pipeline. The length of each stage in the software pipeline is the Initiation Interval (II), which is the rate at which new loop iterations are started. An overlapped lifetime has a live range longer than the II, and as a consequence, the current iteration writes a new value to a register before a previous loop iteration has fin-ished using the old value. Hardware and software solutions for dealing with overlapped lifetimes have been proposed by re-searchers and also implemented in commercial products. These solutions include rotating register files, register queues, modulo variable expansion, and post-scheduling live range splitting. Each of these approaches has drawbacks for embedded systems such as an increase in silicon area, power consumption, and code size.

Effectively controlling a robot remotely, teleoperation, offers benefits in improving human safety, reducing workload, providing location accessibility, and achieving convenient operation. In previous work, teleoperator task performance has been shown to improve with the addition of force feedback. Here we report on a system providing a human operator with a six-axis kinesthetic and visual force display, its implementation and experimental evaluation. A system design containing both a six-axis force reflective hand controller and a stereo video head-mounted display with 3D force display overlay is presented. The system was utilized to evaluate the effects of force display augmentation and substitution by controlling a dexterous anthropomorphic robot called "Robonaut" in the performance of a typical space teleoperation drill task. The human test subject performance under four different force feedback scenarios, no force display, visual force display, kinesthetic force display and bisensory force display, is measured. Experimental results indicate a noticeably lower cumulative force, torque, and variance, lower peak force/torque, yet unchanged task completion times under the presence of visual or kinesthetic force display.

Interaction with medical volume data has often been difficult, due to the large memory and computational power required. By taking advantage of current high-end graphics hardware, we have developed a volumetric virtual environment that provides the ability to help people interact with the volumetric Visible Human data set. The application enables the user to explore the interior of a virtual human body in a natural and intuitive way.

Future computer networks are expected to carry bursty real-time traffic with stringent time-delay requirements. Popular shortest-path routing protocols have the disadvantage of causing bottlenecks due to their single-path routing. We propose a real-time routing and scheduling scheme that randomly distributes the traffic load over all available paths to the destination for load balancing and transmits the packet with the most urgency ahead of the other packets at a switch in a packet-switched network with two objectives-minimizing packet loss in the network and maximizing network throughput. Our scheme consists of two components, a routing algorithm and a scheduling algorithm. The routing algorithm randomly distributes data packets over the whole network to remove bottlenecks caused by the single-path routing of shortest-path routing protocols. The end-to-end delay is bounded by the scheduler, which uses a least-laxity scheduling algorithm. Our simulation results indicate that the proposed scheme decreases the number of packets dropped due to deadline miss and due to buffer overflow in transit in a network and increases network throughput. A desirable by-product is that the traffic load on the network tends to get evenly distributed.

The purpose of routing protocols in a computer network is to maximize network throughput. Popular shortest-path routing protocols have the disadvantage of causing bottlenecks due to their single-path routing. That is, the shortest path between a source and a destination may become highly congested even when many other paths have low utilization. In this paper, we propose a routing scheme for hierarchically structured computer networks such as the Internet. The new hierarchical routing algorithm, which we call hierarchical load-balanced routing (HLBR), is based on distance-vector routing algorithms. HLBR randomly balances traffic load over the whole network; therefore, it removes bottlenecks and increases network throughput. For each data message to be sent from a source s to a destination d, the proposed routing protocol chooses randomly each of three intermediate nodes from a selected set of network nodes, and routes the data message along a path from s through the three intermediate nodes to d. This increases the effective bandwidth between each pair of nodes.

Few studies of object-oriented databases deal with their security, a fundamental aspect of systems with complex data structures. Most authorization systems give users who own resources only some basic control over them; here, we provide users with more direct control over their resources by associating with each grant propagation numbers. Propagation numbers govern the grantability and exercisability of the privileges. Of particular interest in our study of authorization in an OO environment is the combination of inheritance and granting of privileges. Diverse policies are discussed and implemented in a test-bed system.

Radial basis-function networks (RBFNs) have recently attracted interest, because of their advantages over multilayer perceptrons as they are universal approximators but achieve faster convergence since only one layer of weights is required. The least squares method is the most popularly used in estimating the synaptic weights which provides optimal results if the underlying error distribution is Gaussian. However, the generalization performance of the networks deteriorates for realistic noise whose distribution is either unknown or non-Gaussian; in particular, it becomes very bad if outliers are present. In this paper we propose a positive-breakdown learning algorithm for RBFNs by applying the breakdown point approach in robust regression such that any assumptions about or estimation of the error distribution are avoidable. The expense of losing efficiency in the presence of Gaussian noise and the problem of local minima for most robust estimators has also been taken into account. The resulting network is shown to be highly robust and stable against a high fraction of outliers as well as small perturbations. This demonstrates its superiority in controlling bias and variance of estimators.

In this paper, the problem of scheduling independent tasks with bottleneck resource constraints is investigated. There is a set of independent tasks $\mathcal{T} = \{T_{1}, \dotsc , T_{n}\}$ and a set of resources $\mathcal{R} = \{R_{1}, \dotsc , R_{m}\}$. The available amount of resource $R_{j}$, for $j = 1, \dotsc , m$, is (normalized to) 1. Each task $T_{i}$’s execution requires at least $\lambda_{(i,j)} \leq 1$ units of resource $R_{j} \in \mathcal{R}$, and with these minimal resources, the execution time of $T_{i}$, is $\tau_{i}$. If there is a resource $R_{\beta (i)}$ so that $T_{i}$ can use $\xi$ units $(\xi \geq \lambda_{(i,\beta (j))})$ and reduce the execution time to $\frac{{\lambda _{(i,\beta (i))} }}{\xi }\tau _i $, we say that $T_{i}$ can achieve linear speed-up with respect to $R_{\beta (i)}$. It is assumed that for each $T_{i}$, there is one, and only one, resource $R_{\beta (i)}$ that is $T_{i}$’s bottleneck resource: $T_{i}$ can use $\xi$ units of ${R_{(i,\beta (i))} }$ and achieve linear speed-up, where $\xi$ is between $\lambda _{(i,\beta (i))}$ and $\Lambda _{(i,\beta (i))} \leq 1$. The problem is to find a feasible schedule for all the tasks in $\mathcal{T}$ that has a shortest overall makespan. This problem is a combination of two previously studied problems—scheduling tasks with resource constraints [SIAM J. Comput., 4 (1975), pp. 187–200.] scheduling parallel tasks [SIAM J. Comput., 21 (1992), pp. 281–294]. A variant of this problem was studied in [J. Combin. Theory, 21 (1976), pp. 257–298]. Because the problem is NP-hard, we propose the ECT (earliest completion time) algorithm as a heuristic solution and show that the performance ratio of the ECT makespan $M_{\text{ECT}}$ to the optimal makespan $M_{\text{OPT}}$ is bounded by $2l + m + 1$, where l is the number of the bottleneck resources in $\mathcal{R}$. When $l = 0$, this is exactly the performance bound shown by Garey and Graham in [SIAM J. Comput., 4 (1975), pp. 187–200].

An approach for designing high-performance ADT (abstract data type) components for massively parallel systems without sacrificing information hiding is presented. This approach merges information hiding clients and servers to achieve high communication bandwidth for transmitting requests and receiving responses. It uses multi-entry data structures, massive-state-transition interface operations, and a four-level decomposition approach to achieve both structured programming and information hiding within the ADT implementation. To facilitate the systematic design of various ADTs, they have been classified into three classes: unrelated, crystalline, and amorphous collections. The authors present general design decisions for each layer of each class of ADT and illustrate the theory with a detailed example from each class.>

Immutable codes have the property that information recorded with them on write-once memories such as digital optical discs cannot be changed. Write-once memory permits changing a0into a1, but once a1is written it cannot be changed back into a0. Most commonly used codes do not have the property of immutability. After a very general definition of immutability is given, algorithms which test a given code for immutability are developed. Algorithms are presented for fixed-length codes as well as for variable-length codes.

Digital optical disks are interesting as a storage medium because they combine large inexpensive storage capacity with relatively fast access. They also have a property which appears to make them attractive to the security minded, namely they do not permit bits to be rewritten. In other words, a 0 can be changed to a 1, but a 1 cannot be changed back to a 0. In a recent paper, however, Rivest and Shamir show that this does not prevent the changing of information; in fact, at relatively small additional cost (slightly longer codes), information can be changed as often as is desired. In the present paper we are concerned with issues in the integrity of data stored on digital optical disks or any other write-once storage medium, for that matter. A number of practically useful schemes are presented. In particular, immutable codes are discussed, both fixed length and variable length. These are codes which guarantee that information may not be changed. Also introduced are alternative schemes, such as control key schemes, which guarantee either unconditionally or with a certain probability that information cannot be changed if it is stored on digital optical disks. The main distinguishing feature of these schemes is the price to be paid for implementing them, i. e., the number of additional bits required for obtaining the desired protection against subversion.