Search

3D Computer Vision algorithms are a subject of research and application for several industrial processes. The Volume of Interest (VOI) usually refer to sub-objects with basic shapes for computing these algorithms. However, in many cases the objects are available as irregular shapes with many vertices, and in order to apply algorithms effectively, it is essential to compute the largest volume parallelepipedon contained in 3D objects. There are no other approximation algorithms for finding the largest volume parallelepipedon of arbitrary orientation inscribed in a closed 3D contour with a computational cost better than the algorithm proposed in this paper, been $O(n^{3})$ .

The presented article details the modeling of a UMTS phone and its use in specific absorption rate (SAR) and thermal simulations. The obtained thermal results are compared to real-life measurements performed using a commercial thermal imager (Fluke Ti400). The comparison could be beneficial in designing exposure setups where non-thermally induced effects are to be explored, assuring thermally optimal conditions within the total volume of interest and not only on the surface of the exposed object.

Because of the high aggressiveness and lethality of lung cancer, its early detection and accurate characterization are among the most investigated challenges in the last years. Biomedical imaging is an important technology in lung cancer assessment, strongly impacting decision making in clinical practice, also employed as a provider of predictive imaging biomarkers. In this context, radiomics approach, which consists of mining vast arrays of quantitative features derived from digital images, has shown promising application perspectives, but suboptimal standardization and controversial results emerged. This work presents the design of a completely automated pipeline for the non invasive in-vivo characterization of Non-Small Cell Lung Cancer (NSCLC), devised to be a support for radiologists and physicians, and to speed up the diagnostic process. Our pipeline exploits data from routinely acquired PET and CT images in order to automatically obtain a reliable segmentation of the tumor lesion: accurate textural features are computed in the detected Volume of Interest (VOI), thus providing data for the characterization of lung lesion through machine learning algorithms. We evaluated our pipeline on real datasets supplied by a private hospital. Our approach reached a mean accuracy of $94.2\pm 5.0\%$ for the VOI segmentation, and it showed the potential of PET/CT features in differentiating both between primary and metastatic lung lesions and between primary lung cancer subtypes.

We evaluated how magnetic resonance imaging robustness depends on dual-row transmit array geometry, head size, position of a head inside the array and array tuning condition. A clear winner was not found. For overlapped arrays, different tuning conditions resulted in larger peak-to-peak variation for circular polarization excitation, but B1+ (averaged over the volume of interest) decrease was as a bit less for excitations that minimized B1+ root-mean-square inhomogeneity as compared with non-overlapped arrays. For the non-overlapped array, the safety excitation efficiency was better if the distance between the crown of the head and the end of the array was small. Decreasing the distance to the array shield resulted in improved array robustness at the cost of decreased B1+.

We evaluated the performance of several MRI radiofrequency dual-row transmit arrays with a range of overlapping distances, head positions, and tuning conditions. For the geometries, excitation conditions and volume of interest that were investigated, location of the volume of interest in middle of the coil and a small — 10 mm — overlap is preferable. The results provided confirm that with the preferable setup, safety excitation efficiency is practically insensitive to tuning condition, and head position, thus the safety margin can be relative small, allowing shorter scan times.

In this paper, a novel microcalcification detection in Digital Breast Tomosynthesis (DBT) is proposed, which allows detecting subtle microcalcifications that have relatively low contrast in DBT slices. 3D Foveal segmentation is employed for volume of interest (VOI) detection using adaptive thresholding scheme by local volume contrast. Furthermore, in order to conserve structural characteristics of clustered microcalcifications, features from VOI maximum ray-tracing images with three orthogonal directions are extracted. Extensive and comparative experiments were carried out to demonstrate the effectiveness of the proposed 3D Foveal segmentation and the feature extraction method on clinical DBT cases. The results showed that the proposed microcalcification detection achieves a sensitivity of 98% at 0.5 FP per volume.

We present a longitudinal magnetic field gradient-coil with optimized uniformity within the volume of interest (VOI). The accuracy of the optimization algorithm was confronted with measurements of the spatial dependence of the magnetic field within the VOI in a prototype-coil. The proposed device integrates the gradient-unit of a fast-field-cycling magnetic resonance imaging (MRI) apparatus of own design.

A fast detection and diagnosis algorithm is proposed to improve the effectiveness and efficiency of abdominal incisional hernia lesions diagnosis. Firstly, hernia lesions were obtained by detecting black targets in ultrasound images of the coronal plane. Secondly, the morphological method was used to remove false black targets. Thirdly, the volume of interest (VOI) was cropped from the original data set to define the lesions area. Finally, the volume of lesions was measured. Results demonstrated that the proposed algorithm can effectively detect abdominal incisional hernia lesions with automated 3D ultrasound images. It can also provide more diagnostic information by measuring the volume of lesions.

The Cosmic Ray Inspection and Passive Tomography (CRIPT) project is investigating muon scattering tomography (MST) for applications in border security, nuclear non-proliferation, and nuclear waste characterization. The construction of the full-scale prototype MST system began in the Summer of 2011 and was completed in September 2012. The CRIPT detector employs 12 layers of scintillator to track atmospheric muons before and after passage through a volume of interest, and to estimate each muon's momentum. The total height of the system is 5.5 m and its weight is 20 tonnes. Details of its construction are presented. After the integration of the custom data acquisition electronics, the commissioning of the CRIPT detector began. The first tomographic images were obtained in October 2012 and are presented here.

Magnetic fields and their applications in medicine and biology are an active area of research in recent years. Some of these applications include chronic wound healing, electrical stimulation of cardiac tissue and medical imaging among others. The effectiveness of these applications strongly relies on the uniformity of the field, a feature closely related to the arrangement of the coils that generates it. In this paper, genetic algorithms are used to create appropriate arrangements of coils in order to generate a field with a wide volume of uniformity. Results show that the proposed methodology is an effective tool for designing coil arrangements capable of generating magnetic field within an arbitrary volume of interest in which field uniformity is higher than in designs using a traditional approach.

Synthetic aperture processing critically depends on the coherent ping-to-ping nature of the collected data. Phase errors are introduced when the sonar path deviates from a straight line and when the sound speed in the volume of interest is not perfectly constant. Much effort has been spent measuring and estimating corrections to the sonar motion. Phase errors due to inhomogeneous sound speeds are commonly corrected using autofocus algorithms. This paper introduces an autofocus algorithm that constructs an effective correction vector based on a low-order sine series approximation.

We are developing an imaging system that enables the brain of a freely moving animal to be imaged with microPET while simultaneously observing its behaviour. Towards this end, we investigated the feasibility of reconstructing a motion-tracked volume of interest (VOI) in the presence of an extraneous activity compartment with unknown motion. A dual phantom study was performed to simulate movements of a freely moving animal. Both phantoms were moved through discrete positions but only one phantom (representing the head) was tracked. The multiple acquisition frames (MAF) and LOR rebinning methods were applied based on the measured motion of the tracked phantom. We also investigated alternative approaches that are hybrids of these two methods. We found that LOR rebinning causes up to 90% `lost events' (events that would have been measured had motion not occurred) when applied to a freely moving target and this fraction can be significantly reduced using the hybrid approaches, resulting in improved image quality. The MAF-based motion correction yields good results but is not practical for unconstrained motion due to the assumption of no motion within each time segment. We conclude that it is feasible to reconstruct a target VOI in the presence of extraneous activity whose motion is unknown, provided the target motion is accurately tracked.

Pharmacokinetic modeling with dynamic contrast-enhanced MRI (DCE-MRI) has been established as an in vivo technique that estimates the permeability of the vessels. In this study, the most reliable model for permeability assessment of the focused ultrasound (FUS)-induced BBB opening was determined using DCE-MRI. Six mice were sonicated on the right hippocampal region using FUS at the peak-rarefactional acoustic pressure of 0.30 MPa, after intravenous injection of lipid-shelled microbubbles (4–5 μm in diameter). The general kinetic model (GKM) and the reference region model (RRM) were applied on the dynamic MR images to estimate the K trans values. The GKM-derived K trans for a specific volume of interest was found to be 0.02 ± 0.0123 min−1, while RRM yielded statistically insignificantly higher values, 0.03 ± 0.0167 min−1. Both GKM and RRM exhibited significantly low K trans values in the control region (0 and 8.5 ± 12×10−4 min−1 respectively). The two models showed excellent correlation (R2 =0.97), verifying the consistency of the permeability measurements. However, the poor time-concentration curves of RRM suggest that GKM is more suitable for permeability estimation. In conclusion, GKM was found to be the optimal model for the permeability assessment of the FUS-induced BBB opening and the associated drug delivery efficacy.

We address the task of detecting surprising patterns in large textual data streams. These can reveal events in the real world when the data streams are generated by online news media, emails, Twitter feeds, movie subtitles, scientific publications, and more. The volume of interest in such text streams often exceeds human capacity for analysis, such that automatic pattern recognition tools are indispensable. In particular, we are interested in surprising changes in the frequency of n-grams of words, or more generally of symbols from an unlimited alphabet size. Despite the exponentially large number of possible n-grams in the size of the alphabet (which is itself unbounded), we show how these can be detected efficiently. To this end, we rely on a data structure known as a generalised suffix tree, which is additionally annotated with a limited amount of statistical information. Crucially, we show how the generalised suffix tree as well as these statistical annotations can efficiently be updated in an on-line fashion.

This paper reviews coordinated motion control strategies of mobile sensors in mobile surveillance systems. Mobile surveillance systems include a vast array of mobile sensing nodes with varying sensing modalities that can sense continuously the volume of interest. These distributed nodes are capable of sensing, processing, mobilization and communication with other nodes. One of the fundamental problems of mobile surveillance systems is how to coordinate these distributed nodes in such a way that they can move together in concert. Based on the nature of the surveillance task, three coordinated motion control strategies are described. These strategies are direct control, intentional control and emergent control.

Surveillance operations include the timely detection, localization, recognition and identification of objects and events, their relationships, activities, and plans, in a given Volume Of Interest (VOI). Distributed sensors when properly managed can cooperatively provide a complete, accurate, and timely information about the presence and activity of all objects or events within a VOI. This paper presents a holonic sensor management approach that can be used to handle cooperative tasks such as target cueing and handoff in military surveillance operations.

Multimodal tomographic images (e.g., MRI, PET images) provide important information for research, diagnosis and therapy of brain diseases. Structural and functional properties and changes in cortical brain regions are often examined quantitatvely in certain volumes of interest (VOIs) bounded by tissue borders and cortical sulci as important VOI borders. Thus, VOI segmentation is an important prerequisite for this kind of brain analysis. In order to avoid time-consuming slice-by-slice segmentation of VOIs, the new semi-automatic method aims at segmenting cortical VOIs based on 3D brain surface visualization and thus defining VOI borders along cortical sulci much easily than in successive 2D slices. The method was evaluated based on phantoms with simulated sulcus and class properties and real brain data sets. Promising results were obtained regarding segmentation accuracy and computation time.

Micro CT based finite element modeling of trabecular and cortical bone is an effective technique for determining bone's mechanical properties. It is unknown, however, to what degree the selection of the volume of interest by the user via contour lines affects bone morphology and mechanical properties. Therefore four bones were each set up with different contour lines and the results for mechanical and morphological parameters compared. As the results differed for the same bone up to , this study gets to the conclusion that an automatic script for drawing contour lines is the only way to get repeatable and comparable results.

In a Volume of Interest (VOI) based data analysis of Positron Emission Tomography (PET), the relevance of the extracted Time Activity Curves (TACs) is dependent on VOIs delineation accuracy. Several authors proposed unsupervised methods to draw automatically VOIs in PET studies. Among them, Wong proposed a k-means approach [1] and Brankov proposed Expectation-Maximization methods based on the similarity metrics [2], and compared his methods to others [1; 3; 4]. We have designed an automated segmentation method, the Local Means Analysis (LMA), based on local differences in the pharmacokinetics. The method is validated on simulated rat phantom images that include physiological movements, as well as on a real dataset of dynamic PET images acquired in rats.

The accuracy of the estimate of tumor I-131 activity in therapy patients using CT-SPECT fusion is difficult to establish. In our work, we have had the impression that a tumor volume of interest (VoI) from CT sometimes does not perfectly overlay the true position of the tumor in the SPECT image set. However, the magnitude of the difference between the activity estimate from the procedure and that from a perfect overlay is uncertain. Although we have not estimated that magnitude, we have investigated how much the activity estimate increases if we apply a last variation on the fusion to change the location of the tumor VoI so as to increase tumor counts. The results show that the algorithm can be effective in registering tumors in CT and SPECT locally as proved by visual checks of the optimized location of the tumor VoI in SPECT. For 14 tumors in seven patients the increases in tumor counts average 6.65%. The maximum increase is 26.7%.

We are investigating automatic image registration methods that can be used for interventional magnetic resonance imaging (iMRI) guided radiofrequency (RF) thermal ablation of prostate cancer. We tested the ability of slice-to-volume registration between iMRI slice images and high-resolution MRI volumes. Images were acquired from a conventional 1.5 T and an interventional 0.2 T MRI system. We evaluated the registration quality by calculating 3D displacement on a voxel-by-voxel basis over a volume of interest between slice-to-volume registration and volume-to-volume registration that was previously shown to be quite accurate. Visual inspections such as color overlay and contour overlap were also used for registration evaluation. More than 300 registration experiments were performed on MR images of volunteers. Results showed that the registration was quite robust and accurate (

A system is described for 3-d sizing of arteriosclerotic plaques and other vascular structures using ultrasound. The system consists of a conventional high resolution B-mode scanner, a precision scan head position monitor and a pc based evaluation and image processing unit. 3-d images are obtained while scanning the volume of interest where the coordinates of each slice are simultaneously transferred whith the video digitized images into the microcomputer. 3-d structures are outlined by user interactive structure delineating. The system is currently under clinical evaluation.