This paper describes a new airway segmentation algorithm that improves the speed of morphological-based segmentation approaches. Airway segmentation methods based on morphological operators suffer from the indiscriminant application of all operators to a large area. Using the results of three-dimensional (3D) region growing, the discrete application of larger operators is possible. This change can greatly decrease the execution time of the algorithm. This hybrid approach typically runs 5 to 10 times faster than the original algorithm. 3D adaptive region growing, morphological segmentation, and the hybrid approach are then compared via data obtained from human volunteers using a Marconi MX8000 scanner with the lungs held at 85% TLC. Results show that filtering improves robustness of these techniques. The hybrid approach allows for the practical use of morphological operators to create a clinically useful segmentation. We also demonstrate the method's utility for peripheral nodule analysis in a human case.

This paper describes a method to track camera motion of a real endoscope by using epipolar geometry analysis and CT derived virtual endoscopic images. A navigation system for a flexible endoscope guides medical doctors by providing navigation information during endoscope examinations. This paper tries to estimate the motion from an endoscopic video image based on epipolar geometry analysis and image registration between virtual endoscopic (VE) and real endoscopic (RE) images. The method consists of three parts: (a) direct estimation of camera motion by using epipolar geometry analysis, (b) precise estimation by using image registration, and (c) detection of bubble frames for avoiding miss-registration. First we calculate optical flow patterns from two consecutive frames. The camera motion is computed by substituting the obtained flows into the epipolar equations. Then we find the observation parameter of a virtual endoscopy system that generates the most similar endoscopic view to the current RE frame. We execute these processes for all frames of RE videos except for frames where bubbles appear. We applied the proposed method to RE videos of three patients who have CT images. The experimental results show the method can track camera motion for over 500 frames continuously in the best case.

The characterization of the vessel and its lesion is part and parcel of any patient specific planning of endovascular interventions. Thanks to the virtual exploration of a CT volume image acquired under clinical routine conditions we provide the practitioner with the primary elements to plan the intervention. Thus, the curve representing the evolution of the area of the lumen cross-section along the vessel centerline as well as the localization of the stenosis are derived from the virtual exploratory navigation. Due to the lack of reference in medical applications, an animal model of stenosis has been created and used to evaluate our approach. The stenosis characterization through virtual exploratory navigation has been applied to a set of ten sheep iliac arteries and compared with the characterization performed from classical reformatted CT slices and histological cross-sections. The validation of this virtual navigation based analysis is essential to the elaboration of planning systems aimed at the prevention of restenosis phenomena in transluminal angioplasty (accurate localization and covering of the lesion), as well as the treatment of the restenosis phenomena by intravascular brachytherapy (taking the vessel geometry into account).

The purpose of this study was the development of a method for fast and efficient analysis of dynamic MR images of the female breast. The image data sets were acquired with a saturation-recovery turbo-FLASH sequence facilitating the detection of the kinetics of the contrast agent concentration in the whole breast with a high temporal and spatial resolution. In addition, a morphological 3D-FLASH data set was acquired. The dynamic image data sets were analyzed by tracer kinetic modeling in order to describe the physiological processes underlying the contrast enhancement in mathematical terms and thus to enable the estimation of functional tissue specific parameters, reflecting the status of microcirculation. In order to display morphological and functional tissue information simultaneously, we developed a multidimensional visualization system, which enables a practical and intuitive human-computer interface in virtual reality. The quality of real-time volume visualization (using 3D-texture mapping) mainly depends on two factors: number and spatial resolution of the sampling slices. Furthermore, to allow the representation of both the MR signal and the relevant set of model parameters, an adaptation of these quantities to the size of the texture element might be necessary. - Detection and localization of multiple breast lesions may be an important application.

Dual-modality imaging is a technique where computed tomography or magnetic resonance imaging is combined with positron emission tomography or single-photon computed tomography to acquire structural and functional images with an integrated system. The data are acquired during a single procedure with the patient on a table viewed by both detectors to facilitate correlation between the structural and function images. The resulting data can be useful for localization for more specific diagnosis of disease. In addition, the anatomical information can be used to compensate the correlated radionuclide data for physical perturbations such as photon attenuation, scatter radiation, and partial volume errors. Thus, dual-modality imaging provides a priori information that can be used to improve both the visual quality and the quantitative accuracy of the radionuclide images. Dual-modality imaging systems also are being developed for biological research that involves small animals. The small-animal dual-modality systems offer advantages for measurements that currently are performed invasively using autoradiography and tissue sampling. By acquiring the required data noninvasively, dual-modality imaging has the potential to allow serial studies in a single animal, to perform measurements with fewer animals, and to improve the statistical quality of the data.

The quantification of rapid hemodynamic reactions to wide and slow breathing movements has been performed, by two modalities (gamma) -left ventriculography of ^99mTc-labeled blood volume, in anterior oblique incidence on standing and even exercising healthy volunteers and cardiac patients. A highly sensitive stethoscope delivered whole (gamma) -counts acquired at 30 msec intervals in a square field of view including the left ventricle, in a one dimensional low resolution imaging mode for beat to beat analysis. A planar 2D (gamma) -camera imaging of the same cardiac area was then performed without cardiac gating for alternate acquisitions during deep inspiration and deep expiration, completed by a 3D MRI assessment of the stethoscope detection field. Young healthy volunteers displayed wide variations of diastolic times and stroke volumes, as a result of enhanced baroreflex control, together with +/- 16% variations of the stethoscope's background blood volume counts. Any of the components of these responses were shifted, abolished or even inverted as a result of either obesity, hypertension, aging or cardiac pathologies. The assessment of breathing control of the cardiovascular system by the beat to beat (gamma) -ventriculography combined with nuclear 2D and 3D MRI imaging is a kinetic method allowing the detection of functional anomalies in still ambulatory patients.

Inverse electrocardiography has been developing for several years. By coupling electrocardiographic mapping and 3D+time anatomical data, the electrical excitation sequence can be imaged completely noninvasively in the human heart. In this study, a bidomain theory based surface heart model activation time imaging approach was applied to single beat data of atrial and ventricular depolarization. For sinus and paced rhythms, the sites of early activation and the areas with late activation were estimated with sufficient accuracy. In particular for focal arrhythmias, this model-based imaging approach might allow the guidance and evaluation of antiarrhythmic interventions, for instance, in case of catheter ablation or drug therapy.

Discordance between lesion severity from angiocardiography and physiological effects has been reported elsewhere. Quantification of myocardial perfusion during the angiography procedure may supply additional information about short- and long-term outcomes and may be helpful for clinical decision making. In previous works, myocardial perfusion has been assessed using time density curves (TDC), which represent the contrast medium dilution over time in the myocardium. The mean transit time (MTT), derived from the TDC, has been reported as a good indicator of the regional myocardial perfusion. Our objective is to estimate the accuracy and reproducibility of MTT estimation on digital flat panel (DFP) images. We have simulated typical myocardium TDC obtained with a DFP cardiac system (Innova 2000, GE), taking into account scatter and noise. Logarithmic or linear subtractions have been applied to derive a contrast medium concentration proportional quantity from image intensity. A non-linear minimisation realises the model curve fitting. MTT estimates are more stable with linear subtraction in presence of scatter. However logarithmic subtraction presents smaller bias when scatter level is small. Both approaches are equally sensible to image noise. Linear subtraction should be preferred. Image noise has a high influence on MTT accuracy and we may reduce.

There are open questions concerning the hemodynamics during cardiopulmonary resuscitation (CPR). The purpose was to evaluate a model of the blood flow during CPR in specified anatomic regions. After cardiac arrest, one intubated swine under full intensive care supervision was scanned during CPR using an automated resuscitation device. CT scans were performed with an EBCT in the 50ms modus at eight levels, therefore covering most of the heart and pulmonary vessels. 50ml contrast agent was administered with 10ml/sec and a delay of five seconds to visualize the contrast agent passage through the heart and pulmonary circulation. The gray-value changes in previously specified ROIs were directly correlated with the resuscitation device position in respect to the thorax. The effects of CPR on the blood flow could be visualized dynamically by quantifying the contrast enhancement. The increase of gray values could be estimated with different delays, depending on the anatomical situation. The inflow and outflow dependent on thumper dynamics could be estimated. At the onset of contrast medium inflow, turbulence could be visualized in the right ventricle, which are caused by the inhomogeneous contrast medium distribution. Triggered EBCT during CPR offers the opportunity to study regional blood flow depending on chest compression.

The purpose of this paper is to present results of a study comparing CT brain perfusion image calculated using the maximum slope and deconvolution method on data from 32 patients. The 32 patients were organized into two groups. One group was obtained using a 4cc/sec contrast injection rate and 10 sec injection period; the other group was obtained using a 6cc/sec injection rate and 7 sec injection period. All clinical data were analyzed using both the maximum slope and deconvolution methods. Perfusion maps computed from the two methods were reviewed by radiologists. The contrast enhanced CT data were noisy, especially in the white matter area. Our results showed that, for both methods, perfusion maps from 4cc/sec injection were noisier than those using a 6cc/sec injection. However, both 4cc/sec and 6cc/sec produced useful diagnostic images. Our qualitative side-by-side studies also showed that perfusion images from the maximum slope and deconvolution are both clinically useful and substantially equivalent.

This paper presents the experimental setup and preliminary results of a near infrared CCD camera based Photoplethysmography Imaging (PPGI) system, which has been shown to be suitable for contactless and spatially resolved assessment of rhythmical blood volume changes in the skin. To visualize the complex rhythmical patterns in the dermal perfusion the Wavelet Transform is utilized. It is able to jointly assess time and frequency behavior of signals and thus allows to analyze instationary oscillations and variabilities in the different human rhythmics. The presented system is expected to provide new insights into the functional sequences of physiological tissue perfusion as well as of the perfusion status in ulcer formation and wound healing.

The common and important change of pulmonary hemodynamics is represented by increased or decreased pulmonary blood flow (PBF) and increased pulmonary vascular resistance (PVR). We made 3 hemodynamic models in 5 dogs, that is, increased and decreased PBF model and increased PVR model. CT perfusion scan was performed. Perfusion parameters including blood flow (BF), blood volume (BV), mean transit time (MTT), and maximal slope (MS) were analyzed. In normal state, blood flow was affected by gravity and dependent area showed higher BF, BV and lower MS, MTT than non-dependent area. First, decreased PBF model showed no significant change in BV and elongation of MTT. Secondly, increased PBF model showed slightly increased BV and decreased MTT. Thirdly, increased PVR model showed significant decrease of BF, BV, and MS and slight increase of MTT without statistical significance. However, it was noticeable that the distribution of MTT according to gravity in normal lung was completely reversed in increased PVR model. In conclusion, on the basis of our understanding of perfusion characteristic in normal state, we can detect and evaluate the abnormal regional hemodynamic change in lung. Predicting the change of pulmonary vascular resistance should be possible by thorough analysis of CT perfusion parameters.

A rapid multi-phase 3D tagging method based on TrueFISP sequence and its postprocessing algorithm are proposed for 3D regional myocardial motion analysis. SPAMM (SPAtial Modulation of Magnetization) plane tags perpendicular to the read-out direction were imposed for easy tagged-image analysis. To acquire images covering a whole heart within a single breath hold (25 heartbeats), an ellipsoidal cylinder region in the 3D k-space was filled in a 3D centric reordering manner. The postprocessing algorithm was based on a Hessian matrix. Since the eigenvalues of the Hessian matrix defined at each voxel represent the curvature of the 3D image along the corresponding eigenvectors, a group of voxels, at which the largest eigenvalue was positive, much greater than the others and for when the corresponding eigenvector pointed almost along the read-out direction, were selected as a tag-plane. On the resulting images, clear tag contrast can be observed in spite of the small number of k-space lines. Almost all tag planes can be extracted with the algorithm. Those results show our methods are potentially clinically useful.

The problem of detecting and tracking SPAMM tags has been the focus of cardiac MRI in the past few years. Among the many methods proposed to do that, the harmonic phase method (called HARP) has a large potential to enable real-time tracking. We propose two improvements to the original HARP technique to address the problem of magnetic field inhomogeneity and improve tracking. We derive a theoretical model for the problem of field inhomogeneity and propose an extension of the simulated phase evolution rewinding technique (SPHERE) whereby consecutive images are acquired using alternating echo time. Then, we propose a simple and computationally efficient approach that allows the problems of disappearing tag lines to be addressed. In particular, the geodesic horizontal and vertical distances between neighboring corner points in the detected tag lines are computed for each image. Once the correspondence between tag lines is established, the tracking of any point on the wall can be obtained given its relationship to the four corner points surrounding its location. The proposed improvements were implemented and applied to numerical phantoms as well as experimental data obtained for the original HARP technique to demonstrate the potential of the new techniques.

Magnetic resonance tagging is a technique for measuring heart deformations through creation of a stripe grid pattern on cardiac images. Typically, sets of tag surfaces are encoded in the tissue appearing as dark lines on 2D images. In this paper, we present a Maximum A Posteriori (MAP) framework for detecting tag lines using a Markov random field defined on the lattice generated by uniform sampling of B-spline models in 3D and 4D. In the 3D case, MAP estimation is cast for finding the optimal solid for the tag features present in the current image set given an initial solid from the previous frame. The method also allows the parameters of the solid model including the number of knots and the spline order to be adjusted within the same framework. Fitting can start with a solid with less knots and lower spline order, and proceed to one with more knots and/or higher order so as to achieve more accuracy. The optimal solids obtained from 3D tracking for all the frames in the image sequence are considered a 4D B-spline model with linear time interpolation. The framework is then applied to arrive at a 4D B-spline model with higher order time interpolation. The method has been validated with 5 sets of in-vivo data, comprised of a sum total of 882 short-axis (SA) and long-axis (LA) images.

The knowledge about the complex three-dimensional (3D) heart wall motion pattern, particular in the left ventricle, provides valuable information about potential malfunctions, e.g., myocardial ischemia. Nowadays, echocardiography (cardiac ultrasound) is the predominant technique for evaluation of cardiac function. Beside morphology, tissue velocities can be obtained by Doppler techniques (tissue Doppler imaging, TDI). Strain rate imaging (SRI) is a new technique to diagnose heart vitality. It provides information about the contraction ability of the myocardium. Two-dimensional color Doppler echocardiography is still the most important clinical method for estimation of morphology and function. Two-dimensional methods leads to a lack of information due to the three-dimensional overall nature of the heart movement. Due to this complex three-dimensional motion pattern of the heart, the knowledge about velocity and strain rate distribution over the whole ventricle can provide more valuable diagnostic information about motion disorders. For the assessment of intracardiac blood flow three-dimensional color Doppler has already shown its clinical utility. We have developed methods to produce strain rate images by means of 3D tissue Doppler echocardiography. The tissue Doppler and strain rate images can be visualized and quantified by different methods. The methods are integrated into an interactively usable software environment, making them available in clinical everyday life. Our software provides the physician with a valuable tool for diagnosis of heart wall motion.

Quantitative assessment of regional heart motion has the potential to provide diagnostic data for assessment of cardiac malfunction. Local heart motion may be obtained with various medical imaging scanners, so the goal is to provide an imaging modality-independent display/analysis technique. In this study, 3D reconstructions of a canine heart before and after infarction were obtained from the Dynamic Spatial Reconstructor (DSR) at 15 time points throughout one cardiac cycle. Deformable models of each time point were created. Through this process regional excursions and velocities in the mesh can be assigned to represent a piece of endocardium, which can be calculated for each time-point interval. These calculations are based on the distance change between a single vertex of the mesh and the model centerline from LV apex to aortic/mitral valve separation. This allows computation of color maps corresponding to regional values of contraction or dilation motion of the endocardium relative to the LV long axis (centerline) during systole and/or diastole. These color maps can be illustrated through model animations and multi view static images. Using functional parametric mappings of disturbances in regional contractility and relaxation facilitates appreciation of the effect of altered structure-to-function relationships in the myocardium.

Estimation of left ventricle motion and deformation from series of images has been an area of attention in the medical image analysis and still remains an open and challenging problem. Left ventricle contractile abnormalities can be an important manifestation of coronary artery disease. The proper motion tracking of left ventricle wall can contribute to isolate the location and extent of ischemic or infarcted myocardium and constitutes a fundamental goal of image modalities, such as Nuclear Medicine. This work describes a method to automatically estimate the velocity vector field for a beating heart based on the study of variation in frequency content in a series of 2D images as time varies. The frequency analysis is performed by computing the Wigner-Ville and the Choi-Williams distributions to each image pixel, yielding the corresponding 3D-frequency spectrum. From this 3D spectrum the local velocity of each pixel is calculated by employing a multiple linear regression model. Experimental validation was carried out using synthetic phantoms that simulate translation and rotation between successive frames. Results obtained from gated SPECT perfusion studies are also presented.

Purpose: To evaluate a filter method to extract noise from 20mAs Computed Tomography (CT) data for virtual colonoscopy screening. Method: Nonlinear Gaussian filter chains (NLGF) applied to CT datasets were used to extract noise. To test the efficiency of NLGF a simulation of different ellipsoidal shells with different levels of noise were used. Phantom studies were performed using a multidetector CT (tube currents 10 to 140mAs). 15 patients at high risk for colon cancer underwent a virtual colonoscopy examination (140mAs) and conventional colonoscopy. Different noise levels were added to each CT raw dataset (analog to 40 and 20mAs scans). Virtual endoscopic fly-throughs were performed and rated by two radiologists (image quality). Results: NLGF were able to extract image noise while preserving image structures down to signal--to--noise ratio levels of 0.5. The phantom studies (perspex bars, simulated polyps) were reconstructed without relevant changes between 20 and 140mAs. There were no significant differences between the endoscopic fly-throughs of 140 and 20mAs examinations (2 readers). Conclusion: NLGF is a promising preprocessing method for effective noise reduction in CT datasets. Edges are preserved as well as accentuated in high contrast images.

We have developed a method that automatically displays, places in sized order and allows viewing of the areas of the colon surface not visualized during initial endoscopic navigational viewing. While complete surface visualization is possible, we demonstrate that all of these missed patches do not have to be reviewed to detect clinically significant colon polyps. CT scans are performed on 147 patients and volunteers after bowel preparation and colon distention with CO₂. After automatic segmentation and electronic cleansing of the colon lumen, the medial axis (centerline) is extracted. Volume rendering fly-through along the centerline is performed and visualized surfaces are marked. To simulate optical colonoscopy, the virtual camera is passed in the antegrade direction. For virtual colonoscopy, the camera is passed both antegrade and retrograde, and the combined visible surface voxel count is recorded. After both fly-throughs, the total visualized surface is recorded and all 'patches' of connected surface area not yet seen are identified, measured, sorted by size, and counted. Clinically significant patches, defined as smallest diameter being > 5mm, are sequentially visualized by stepping through the sorted list until reaching the patch diameter of 5 mm.. for each. By enabling endoscopic navigation in both antegrade and retrograde directions, virtual colonoscopy is able to evaluate behind haustral folds and around sharp bends, thereby visualizing significantly more surface area than optical colonoscopy. Furthermore, automatic marking of the visualized surface area and identifying and viewing unseen patches allows examination of all clinically significant surfaces of the colon.

A multi-network decision classification scheme for colonic polyp detection is presented. The approach is based on the results of voting over several neural networks using different variable sets of size N which are selected randomly or by an expert from a general variable set of size M. Detection of colonic polyps is complicated by a large variety of polypoid looking shapes (haustral folds, leftover stool) on the colon surface. Using various shape and curvature characteristics, intensity, size measurements and texture features to distinguish real polyps from false positives leads to an intricate classification problem. We used 17 features including region density, Gaussian and average curvature and sphericity, lesion size, colon wall thickness, and their means and standard deviations in the vicinity of the prospective polyp. Selection of the most important parameters to reduce a feature set to acceptable size is a generally unsolved problem. The method suggested in this paper uses a collection of subsets of variables. These sets of variables are weighted by their effectiveness. The effectiveness cost function is calculated on the basis of the training and test sample mis-classification rates obtained by the training neural net with the given variable set. The final decision is based on the majority vote across the networks generated using the variable subsets, and takes into account the weighted votes of all nets. This method reduces the flst positive rate by a factor of 1.7 compared to single net decisions. The overall sensitivity and specificity rates reached are 100% and 95% correspondingly. Best specificity and sensitivity rates were reached using back propagation neural nets with one hidden layer trained with the Levenberg-Marquardt algorithm. Ten-fold cross-validation is used to better estimate the true error rates.

To facilitate early detection and removal of colonic polyps, we have developed a method to present the whole colon surface in a single 2D image. To improve detection sensitivity and lower false positives while using the 2D image, we utilize curvature-vector pair and curvature-images to construct a candidate-point image to indicate possible locations of polyps in the 2D image. Planes perpendicular to the centerline are generated. On each plane, equiangularly spaced radial rays are generated, which intersect the colon surface, generating sample points. The centerline index of the planes and the angular index of the radial rays on each plane define the x and y coordinates of a 2D array, respectively. At each sample point, four curves are constructed along the x, the y, and the two diagonal directions, for which four curvatures are calculated. The curvature vectors which pointing into the lumen are set to zero. Then, at each sample point, the values of the maximum and minimum curvature are assigned to the corresponding element in the 2D array. Using this 2D array and a two-step thresholding method, a so-called candidate-point image, which indicates possible locations of polyps, is generated. In the two-step thresholding method, the thresholds are determined by using a five training colon-data sets. This candidate-point image can be used in conjunction with other analyses in polyp detection.

We developed a novel automated technique for segmenting the colonic wall in computer-aided detection of polyps in CT colonography. The technique is designed to minimize the presence of extracolonic components, such as small bowel, in the segmented colon. The colon segmentations were evaluated subjectively by four radiologists. On average, 98% of the visible colonic wall was covered by the segmentation. The amount of extracolonic components was reduced by 50% compared with our previously used anatomy-oriented colon segmentation technique, but approximately 10-15% of the segmentation still contained extracolonic components. When the technique was used with our fully automated computer-aided polyp detection scheme at a 100% by-patient detection sensitivity, the false-positive rate was reduced by 20% from 2.5 false positives to 2.0 false positives per patient. These preliminary results suggest that our new colon segmentation technique can improve the specificity of our CAD scheme significantly without degradation in the detection sensitivity.

To support ophthalmologists in their routine and enable the quantitative assessment of vascular changes in color fundus photographs a multi-resolution approach was developed which segments the vessel tree efficiently and precisely in digital images of the retina. The algorithm starts at seed points, found in a preprocessing step and then follows the vessel, iteratively adjusting the direction of the search, and finding the center line of the vessels. As an addition, vessel branches and crossings are detected and stored in detailed lists. Every iteration of the Directional Smoothing Based (DSB) tracking process starts at a given point in the middle of a vessel. First rectangular windows for several directions in a neighborhood of this point are smoothed in the assumed direction of the vessel. The window, that results in the best contrast is then said to have the true direction of the vessel. The center point is moved into that direction 1/8th of the vessel width, and the algorithm continues with the next iteration. The vessel branch and crossing detection uses a list with unique vessel segment IDs and branch point IDs. During the tracking, when another vessel is crossed, the tracking is stopped. The newly traced vessel segment is stored in the vessel segment list, and the vessel, that had been traced before is broken up at the crossing- or branch point, and is stored as two different vessel segments. This approach has several advantages: - With directional smoothing, noise is eliminated, while the edges of the vessels are kept. - DSB works on high resolution images (3000 x 2000 pixel) as well as on low-resolution images (900 x 600 pixel), because a large area of the vessel is used to find the vessel direction - For the detection of venous beading the vessel width is measured for every step of the traced vessel. - With the lists of branch- and crossing points, we get a network of connected vessel segments, that can be used for further processing the retinal vessel tree.

The purpose of this work is to build a computerized system for the delineation of upper airway structures via MRI and to evaluate its effectiveness for routine clinical use in aiding diagnosis of upper airway disorders in children. We use two MRI protocols, axial T1 and T2, to gather information about different aspects of the airway and its surrounding soft tissue structures including adenoid, tonsils, tongue and soft palate. These images are processed and segmented to compute the architectural parameters of the airway such as its surface description, volume, central (medial) line, and cross-sectional areas at planes orthogonal to the central line. We have built a software package based on 3DVIEWNIX and running on a 450 MHz Pentium PC under Linux system (and on a Sun workstation under Unix) for the various operations of visualization, segmentation, registration, prefiltering, interpolation, standardization, and quantitative analysis of the airway. The system has been tested utilizing 40 patient studies. For every study, the system segmented and displayed a smooth 3D rendition of the airway, its central line and a plot of the cross-sectional area of the airway orthogonal to the central line as a function of the distance from one end of the central line. The tests indicate 97% precision and accuracy for segmentation. The mean time taken per study is about 4 minutes for the airway. This includes operator interaction time and processing time. This method provides a robust and fast means of assessing the airway size, shape, and places of restriction, as well as providing a structural data set suitable for use in modeling studies of airflow and mechanics.

Carotid vessel ultrasound imaging is a reliable non-invasive technique to measure the arterial morphology. Vessel diameter, intima-media thickness (IMT) of the far wall and plaque presence can be reliably determined using B-mode ultrasound. In this paper we describe a semi-automatic approach to measure artery diameter and IMT based on an active contour technique improved by a multiresolution analysis. The operator selects a region-of-interest (ROI) in a series of carotid images obtained from B-mode ultrasound. This set of images is convolved with the corresponding partial derivatives of the Gaussian filter. The filter response is used to compute a 2D gradient magnitude image in order to refine the vessel's boundaries. Using an active contour technique the vessel's border is determined automatically. The near wall media-adventitia (NWMA), far wall media-adventitia (FWMA) and far wall lumen-intima (FWLI) borders are obtained by a least-square fitting of the active contours result. The distance between NWMA and FWLI (vessel diameter) and between FWLI and FWMA (far wall intima-media thickness) are obtained for all images and the mean value is computed during systole and diastole. The proposed method is a reliable and reproducible way of assessing the vessel diameter and far wall intima-media thickness of the carotid artery.

Quantitative analysis of the branching geometry of multiple branching-order vascular trees from 3D micro-CT data requires an efficient segmentation algorithm that leads to a consistent, accurate representation of the tree structure. To explore different segmentation techniques, we use isotropic micro-CT-images of intact rat coronary, pulmonary and hepatic opacified arterial trees with cubic voxel-side length of 5-20 micrometer. We implemented an active topology adaptive surface model for segmentation and compared the results from this algorithm with segmentations of the same image data using conventional segmentation methods. Because of the modulation transfer function of the micro-CT scanner, thresholding and region growing techniques usually underestimate small, or overestimate large, vessel diameters depending on the chosen grayscale thresholds. Furthermore, these approaches lack the robustness needed to overcome the effects of typical imaging artifacts, such as image noise at the vessel surfaces, which tend to propagate errors in the analysis of the tree due to its hierarchical nature. Our adaptable surface models include local gray- scale statistics, object boundary and object size information into the segmentation algorithm, thus leading to a higher stability and accuracy of the segmentation process. 5-20 micrometer. We implemented an active topology adaptive surface model for segmentation and compared the results from this algorithm with segmentations of the same image data using conventional segmentation methods. Because of the modulation transfer function of the micro-CT scanner, thresholding and region growing techniques usually underestimate small, or overestimate large, vessel diameters depending on the chosen grayscale thresholds. Furthermore, these approaches lack the robustness needed to overcome the e*ects of typical imaging artifacts, such as image noise at the vessel surfaces, which tend to propagate errors in the analysis of the tree due to its hierarchical nature. Our adaptable surface models include local gray-scale statistics, object boundary and object size information into the segmentation algorithm, thus leading to a higher stability and accuracy of the segmentation process.

In the study of pulmonary vascular remodeling, much can be learned from observing the morphological changes undergone in the pulmonary arteries of the rat lung when exposed to chronic hypoxia or other challenges which elicit a remodeling response. Remodeling effects include thickening of vessel walls, and loss of wall compliance. Morphometric data can be used to localize the hemodynamic and functional consequences. We developed a CT imaging method for measuring the pulmonary arterial tree over a range of pressures in rat lungs. X-ray micro-focal isotropic volumetric imaging of the arterial tree in the intact rat lung provides detailed information on the size, shape and mechanical properties of the arterial network. In this study, we investigate the changes in arterial volume with step changes in pressure for both normoxic and hypoxic Fawn-Hooded (FH) rats. We show that FH rats exposed to hypoxia tend to have reduced arterial volume changes for the same preload when compared to FH controls. A secondary objective of this work is to quantify various phenotypes to better understand the genetic contribution of vascular remodeling in the lungs. This volume estimation method shows promise in high throughput phenotyping, distinguishing differences in the pulmonary hypertensive rat model.

The combination of quantitative coronary analysis and flow reserve measurements enables the clinician to determine whether a coronary artery stenosis is significant and therefore has to be treated. 2-D SPECT polar diagrams are made to get information on cardiac perfusion. However, no real 3-D comparison between the anatomical coronary angiography data and the perfusion information can be made. In this feasibility study a first approach is made to create fusion images in 3-D of angiograms and SPECT data. From biplane coronary arteriograms (CAGs), both left and right coronary arteries of five patients have been reconstructed as 3-D models. The reconstruction output was automatically converted into Virtual Reality Markup Language (VRML) scenes. The 2-D polar SPECT data were mapped onto a half-ellipsoid and added to the VRML scene. Registration of the three models was performed interactively using VRML and common Internet browsers.

We have developed a waveform shape model-based algorithm for the extraction of blood flow from dynamic arterial x-ray angiographic images. We have carried out in-vitro validation of this technique. A pulsatile physiological blood flow circuit was constructed using an anthropomorphic cerebral vascular phantom to simulate the cerebral arterial circulation with whole blood as the fluid. Instantaneous recording of flow from an electromagnetic flow meter (EMF) provided the gold standard measurement. Biplane dynamic digital x-ray images of the vascular phantom with injection of contrast medium were acquired at 25 fps using a PC frame capture card with calibration using a Perspex cube. Principal component analysis was used to construct a shape model by collecting 434 flow waveforms from the EMF under varying flow conditions. Blood flow waveforms were calculated from the angiographic data by using our previous concentration-distance curve matching (ORG) algorithm and by using the new model-based (MB) algorithm. Both instantaneous and mean flow values calculated using the MB algorithm showed greater correlation, less bias, and lower variability than those calculated using the ORG algorithm when compared to the EMF values. We have successfully demonstrated that use of a priori waveform shape information can improve flow measurements from dynamic x-ray angiograms.

A methodology to construct patient-specific, anatomically and physiologically realistic finite element models of blood flows in stenosed carotid arteries is presented. Anatomical models of carotid arteries with stenosis are reconstructed from contrast-enhanced magnetic resonance angiography (MRA) images using a tubular deformable model along each arterial branch. A surface-merging algorithm is used to create a watertight model of the carotid bifurcation for subsequent finite element grid generation. A fully implicit scheme is used to solve the incompressible Navier-Stokes equations on unstructured grids in three-dimensions. Physiologic boundary conditions are derived from cine phase-contrast MRA flow velocity measurements at two locations below and above the bifurcation. The methodology was tested on image data of a patient with carotid artery stenosis. A finite element grid was successfully generated from contrast-enhanced MRA images, and pulsatile blood flow visualizations were produced. Visualizations of the wall shear stress distribution and of changes in both its magnitude and direction were produced. These quantities may become important in order to characterize healthy and diseased flow and wall shear stress patterns. We conclude that MRA can be used to obtain all the anatomical and physiologic data necessary for realistic modeling of blood flows in carotid arteries with stenosis. Our results confirm that image-based computational fluid dynamics techniques can be applied to the modeling of hemodynamics in carotid arteries with stenosis. These capabilities may be used to advance our understanding of the generation and progression of vascular disease, and may eventually allow physicians to enhance current image-based diagnosis, and to predict and evaluate the outcome of interventional procedures non-invasively.

We've developed a Multi-slice Spiral CT Simulator modeling the acquisition process of a real tomograph over a 4-dimensional phantom (4D MCAT) of the human thorax. The simulator allows us to visually characterize artifacts due to insufficient temporal sampling and a priori evaluate the quality of the images obtained in cardio-pulmonary studies (both with single-/multi-slice and ECG gated acquisition processes). The simulating environment allows both for conventional and spiral scanning modes and includes a model of noise in the acquisition process. In case of spiral scanning, reconstruction facilities include longitudinal interpolation methods (360LI and 180LI both for single and multi-slice). Then, the reconstruction of the section is performed through FBP. The reconstructed images/volumes are affected by distortion due to insufficient temporal sampling of the moving object. The developed simulating environment allows us to investigate the nature of the distortion characterizing it qualitatively and quantitatively (using, for example, Herman's measures). Much of our work is focused on the determination of adequate temporal sampling and sinogram regularization techniques. At the moment, the simulator model is limited to the case of multi-slice tomograph, being planned as a next step of development the extension to cone beam or area detectors.

Accurate reconstruction of the human brain in MRI-T1 images is valuable and important to clinical needs. In this paper, the morphology and snake techniques are proposed to reconstruct a human brain model. First step in our method is to preprocess the volumetric image to remove skull, muscle, fat, and other non-brain tissue. We use a method of 3-d region growing. It has the advantage over thresholding that the resulting objects will be spatially connected, since brain has the connected property. Second, we use clustering method, and than use them to produce an initial estimate of the cortical surface. Third, we propose a novel active contour algorithm to move the snake toward the cortex. Thus we can use the snake to segment the brain. We use a wavelet method to model the external force that significantly increases the capture range of a traditional snake. Afterwards, we render the volumetric image to display the brain from multiple views. Both simulated data and patient data have been use to test the proposed techniques. The proposed method combines various techniques of 3-D morphology, clustering, active contour, wavelet, and volume rendering to accurately, robustly, and automatically reconstruct brain from MRI-T1 images.

CT can be used to study pulmonary structure-function relationships. There is a growing clinical need to match pulmonary structures across individuals to detect abnormal structure due to disease and to compare regional pulmonary function. In this paper, we propose a novel scheme for registering and warping 3-D pulmonary CT images of different subjects in two main steps: 1) identify a set of reproducible feature points for each CT image to establish correspondences across subjects; 2) use a landmark and intensity-based consistent image registration algorithm to warp a template image volume to the rest of the lung volumes. Effectiveness of the proposed scheme is evaluated and visualized using both gray-level and segmented CT images. Results show that the proposed scheme is able to reduce landmark registration error and relative volume overlapping error from 10.5 mm and 0.70 before registration to 0.4 mm and 0.11, respectively. The proposed scheme can be used to construct a computerized human lung model (or atlas) to help detect abnormal lung structural changes.

We compared tomographic scintimammography performed using single photon emission computed tomography (SPECT), positron emission coincidence imaging (PECI) and positron emission tomography (PET). A female thorax phantom was used. Activities of the myocardium, thorax and breasts were adjusted to emulate the count rate observed with patients. Hollow plastic spheres, imitating hot lesions (1.5-20ml), filled with radioactive saline were inserted in the center of each breast. Specific activities of internal organs were adjusted to emulate the count rate observed with patients. SPECT data were acquired with Tc-99m using gamma cameras with NaI(Tl) detectors. A modified FBP (CODE) reconstruction algorithm was used to render SPECT tomographic images. PECI (Siemens E.CAM with NaI(Tl)) and PET (GE Advance with BGO) data were acquired using F-18 FDG. Vendor supplied reconstruction algorithms were used. The reconstructed hot lesions contrast and resolution were investigated. Image quality obtained can be ranked as follows: (1) PET(BGO), (2) PECI(NaI), (3) SPECT(NaI) In conclusion, assuming comparable uptake values of Tc-99m-sestamibi and F-18 FDG, PET seems to be a superior methodology in visualization of breast lesion as compared to SPECT and PECI. All these tomographic methods appear to be promising adjunct to x-ray mammography in difficult to interpret cases.

A bronchoscope can be used to examine the mucosal surface of the airways for abnormalities associated with a variety of lung diseases. The diagnosis of these abnormalities through the process of bronchoscopy is based, in part, on changes in airway wall color. Therefore it is important to characterize the normal color inside the airways. We propose a standardized method to calibrate the bronchoscopic imaging system and to tabulate the normal colors of the airway. Our imaging system consists of a Pentium PC and video frame grabber, coupled with a true color bronchoscope. The calibration procedure uses 24 standard color patches. Images of these color patches at three different distances (1, 1.5, and 2 cm) were acquired using the bronchoscope in a darkened room, to assess repeatability and sensitivity to illumination. The images from the bronchoscope are in a device-dependent Red-Green-Blue (RGB) color space, which was converted to a tri-stimulus image and then into a device-independent color space sRGB image by a fixed polynomial transformation. Images were acquired from five normal human volunteer subjects, two cystic fibrosis (CF) patients and one normal heavy smoker subject. The hue and saturation values of regions within the normal airway were tabulated and these values were compared with the values obtained from regions within the airways of the CF patients and the normal heavy smoker. Repeated measurements of the same region in the airways showed no measurable change in hue or saturation.

A new method is proposed for activation detection in event-related functional magnetic resonance imaging (fMRI). The method is based on nonparametric analysis of selected resolution levels (a subspace) in translation invariant wavelet transform (TIWT) domain. Using a priori knowledge about the activation signal and trends, we analyze their power in different resolution levels in TIWT domain and select an optimal set of resolution levels. A nonparametric randomization method is then applied in the wavelet domain for activation detection. This approach suppresses the effects of trends and enhances the detection sensitivity. In addition, since TIWT is insensitive to signal translations, the power analysis is robust with respect to signal shifts. Nonparametric randomization alleviates the need for assumptions about fMRI noise. The method has been applied to simulated and experimental fMRI datasets. Comparisons have been made between the results of the proposed method, a similar method in the time domain, and the cross-correlation method. The proposed method has shown superior sensitivity compared to the other methods.

Among the methods proposed for the analysis of functional MR we have previously introduced a model-independent analysis based on the self-organizing map (SOM) neural network technique. The SOM neural network can be trained to identify the temporal patterns in voxel time-series of individual functional MRI (fMRI) experiments. The separated classes consist of activation, deactivation and baseline patterns corresponding to the task-paradigm. While the classification capability of the SOM is not only based on the distinctness of the patterns themselves but also on their frequency of occurrence in the training set, a weighting or selection of voxels of interest should be considered prior to the training of the neural network to improve pattern learning. Weighting of interesting voxels by means of autocorrelation or F-test significance levels has been used successfully, but still a large number of baseline voxels is included in the training. The purpose of this approach is to avoid the inclusion of these voxels by using three different levels of segmentation and mapping from Talairach space: (1) voxel partitions at the lobe level, (2) voxel partitions at the gyrus level and (3) voxel partitions at the cell level (Brodmann areas). The results of the SOM classification based on these mapping levels in comparison to training with all brain voxels are presented in this paper.

We propose a technique that enables robust use of blind source separation techniques in fMRI data analysis. The fMRI temporal signal is modeled as the summation of the true activation signal, a physiological baseline fluctuation component, and a random noise component. A preprocessing denoising is used to reduce the dimensionality of the random noise component in this mixture before applying the principal/independent component analysis (PCA/ICA) methods. The set of denoised time courses from a localized region are utilized to capture the region-specific activation patterns. We show a significant improvement in the convergence properties of the ICA iteration when the denoised time courses are used. We also demonstrate the advantage of using ICA over PCA to separate components due to physiological signals from those corresponding to actual activation. Moreover, we propose the use of ICA to analyze the magnitude of the Fourier domain of the time courses. This allows ICA to group signals with similar patterns and different delays together, which makes the iteration even more efficient. The proposed technique is verified using computer simulations as well as actual data from a healthy human volunteer. The results confirm the robustness of the new strategy and demonstrate its value for clinical use.

The objective of this study is to introduce a simple mathematical model for the brain image under both resting and activated states to facilitate both the understanding of the underlying neurophysiology and the realization of data sets. Two data sets were composed to simulate fMRI data. First set consists of a small spot that simulates the activated region superimposed on real baseline data. To simulate the signal enhancement, the hemodynamic response vector multiplies all pixels in the activated spot. Therefore, the resultant spots were added sequentially to the baseline images to create the first data set. The second set was formed by using the proposed model. The model took into account both random and physiological noise that are found in fMRI data. The random noise was assumed to vary from one frame to another while the physiological pattern was assumed of similar pattern throughout the brain with smooth spatial variations. A threshold cross-correlation technique was used on both data sets to compare the resultant activation maps. A falsehood measure was proposed and used as to test the accuracy of the activation detection. Finally, the results between the two data sets are compared to demonstrate the accuracy of the model.

This research builds on our hypothesis that white matter damage and associated neurocognitive symptoms, in children treated for cancer with cranial spinal irradiation, spans a continuum of severity that can be reliably probed using non-invasive MR technology. Quantitative volumetric assessments of MR imaging and psychological assessments were obtained in 40 long-term survivors of malignant brain tumors treated with cranial irradiation. Neurocognitive assessments included a test of intellect (Wechsler Intelligence Test for Children, Wechsler Adult Intelligence Scale), attention (Conner's Continuous Performance Test), and memory (California Verbal Learning Test). One-sample t-tests were conducted to evaluate test performance of survivors against age-adjusted scores from the test norms; these analyses revealed significant impairments in all apriori selected measures of intelligence, attention, and memory. Partial correlation analyses were performed to assess the relationships between brain tissues volumes (normal appearing white matter (NAWM), gray matter, and CSF) and neurocognitive function. Global intelligence (r = 0.32, p = 0.05) and global attentional (r = 0.49, p < 0.01) were significantly positively correlated with NAWM volumes, whereas global memory was significantly positively correlated with overall brain parenchyma (r = 0.38, p = 0.04). We conclude that quantitative assessment of MR examinations in survivors of childhood cancer treated with cranial irradiation reveal that loss of NAWM is associated with decreased intellectual and attentional deficits, whereas overall parenchyma loss, as reflected by increased CSF and decreased white matter, is associated with memory-related deficits.

Virtual colonoscopy provides a safe, minimal-invasive approach to detect colonic polyps using medical imaging and computer graphics technologies. Residual stool and fluid are problematic for optimal viewing of the colonic mucosa. Electronic cleansing techniques combining bowel preparation, oral contrast agents, and image segmentation were developed to extract the colon lumen from computed tomography (CT) images of the colon. In this paper, we present a new electronic colon cleansing technology, which employs a hidden Markov random filed (MRF) model to integrate the neighborhood information for overcoming the non-uniformity problems within the tagged stool/fluid region. Prior to obtaining CT images, the patient undergoes a bowel preparation. A statistical method for maximum a posterior probability (MAP) was developed to identify the enhanced regions of residual stool/fluid. The method utilizes a hidden MRF Gibbs model to integrate the spatial information into the Expectation Maximization (EM) model-fitting MAP algorithm. The algorithm estimates the model parameters and segments the voxels iteratively in an interleaved manner, converging to a solution where the model parameters and voxel labels are stabilized within a specified criterion. Experimental results are promising.

We present an electronic colon cleansing algorithm using a new segmentation technique based on segmentation rays. These rays are specially designed to analyze the intensity profile as they traverse through the dataset. When this intensity profile matches any of the pre-defined profiles, the rays perform certain task of reconstruction. We use these rays to detect the intersection between air and residual fluid, and between residual fluid and soft-tissue. One of the most important advantages of segmentation rays over other segmentation techniques is the detection of partial volume regions. Segmentation rays can accurately detect partial volume regions and remove them if needed. Once partial volume is eliminated, removal of other unwanted regions (e.g., tagged fluid) is relatively easy. This approach to electronic cleansing is extremely fast as it requires minimal computation.

Internet addicted patients (IAPs) have widely been increased, as Internet games are becoming very popular in daily life. The purpose of this study was to investigate regional brain activation patterns associated with excessive use of Internet games in adolescents. Six normal controls (NCs) and eight IAPs who were classified as addiction group by adapted version of DSM-IV for pathologic gambling were participated. 18F-FDG PET studies were performed for all adolescents at their rest and activated condition after 20 minutes of each subject's favorite Internet game. To investigate quantitative metabolic differences in both groups, all possible combinations of group comparison were carried out using Statistical Parametric Mapping (SPM 99). Regional brain activation foci were identified on Talairach coordinate. SPM results showed increased metabolic activation in occipital lobes for both groups. Higher metabolisms were seen at resting condition in IAPs than that of in NCs. In comparison to both groups, IAPs showed different patterns of regional brain metabolic activation compared with that of NCs. It suggests that addictive use of Internet games may result in functional alteration of developing brain in adolescents.

The size and location of the cochlea and cochlear nerve are needed to assess the feasibility of cochlea implantation, provide information for surgical planning, and aid in construction of cochlear models. Models of implant stimulation incorporating anatomical and physiological information are likely to provide a better understanding of the biophysics of information transferred with cochlear implants and aid in electrode design and arrangement on cochlear implants. Until recently MR did not provide the necessary image resolution and suffered from long acquisition times. The purpose of this study was to optimize both Fast Spin Echo (FSE) and Steady State Free Precession (FIESTA) imaging scan parameters for the inner ear and comparatively examine both for improved image quality and increased spatial resolution. Image quality was determined by two primary measurements, signal to noise ratio (SNR), and image sharpness. Optimized parameters for FSE were 120ms, 3000ms, 64, and 32.25kHz for the TE, TR, echo train length, and bandwidth, respectively. FIESTA parameters were optimized to 2.7, 5.5ms, 70 degree(s), and 62.5kHz, for TE, TR, flip angle, and bandwidth, respectively. While both had the same in-plane spatial resolution, 0.625mm, FIESTA data shows higher SNR per acquisition time and better edge sharpness.

In clinically, structural image based brain tissue segmentation as a preprocess plays an important and essential role on a number of image preprocessing, such as image visualization, object recognition, image registration, and so forth. However, when we need to classify the tissues according to their physiological functions, those strategies are not satisfactory. In this study, we incorporated both tissue time-activity curves (TACs) and derived kinetic parametric curves (KPCs) information to segment brain tissues, such as striatum, gray and white matters, in dynamic FDOPA-PET studies. Four common clustering techniques, K-mean (KM), Fuzzy C-mean (FCM), Isodata (ISO), Markov Random Fields (MRF), and our method were compared to evaluate its precision. The results show 41% and 48% less mean errors in mean difference for KPCs and TACs, respectively, than other methods. Combined KPCs and TACs based clustering method provide the ability to define brain structure effectively.

What is the impact of the spin history and position history on signal intensity after the alignment of acquired volumes? This question arises in many fMRI studies. We will focus on spin-history artefacts generated by the position-history of the scanned object. In fMRI an object is driven to steady state by applying a few dummy scans at the start of each measurement. A change in object position will disrupt the tissue's steady state magnetization. The disruption will propagate to the next few acquired volumes until a steady state is reached again. The variables which are involved in changing the longitudinal magnetization are: the shape and the position of the slice profiles, the times at which RF pulses occurred, the equilibrium magnetization map and the T₁ map. Knowledge of these variables enables the prediction of those situations and the locations where the spin-history may compromise the fMRI analysis. In this paper we present a simulation of spin-history artefacts. The simulation shows that these effects, following a displacement, are similar to the transient period at the beginning of the measurement. Introducing gaps between the acquired slices increases these artefacts.

An efficient algorithm for generation of the task reference function has been developed that allows real-time statistical analysis of fMRI data, within the framework of the general linear model, for experiments with event-related stimulus designs. By leveraging time-stamped data collection in the Input/Output time-aWare Architecture (I/OWA), we detect the onset time of a stimulus as it is delivered to a subject. A dynamically updated list of detected stimulus event times is maintained in shared memory as a data stream and delivered as input to a real-time convolution algorithm. As each image is acquired from the MR scanner, the time-stamp of its acquisition is delivered via a second dynamically updated stream to the convolution algorithm, where a running convolution of the events with an estimated hemodynamic response function is computed at the image acquisition time and written to a third stream in memory. Output is interpreted as the activation reference function and treated as the covariate of interest in the I/OWA implementation of the general linear model. Statistical parametric maps are computed and displayed to the I/OWA user interface in less than the time between successive image acquisitions.

There is currently no accurate method to measure airway dimensions on multidetector row computed tomography (multi-slice CT). We developed CT image analysis software to measure airway lumenal area (Ai) and airway wall area (Aaw) and compared these with quantitative morphology of excised human lungs as the gold standard. Airways identified on the CT images (1.25 mm collimation) were matched to airways identified on the lung's cut surface and Ai and Aaw were measured using custom software. The measured morphological airway lumen ranged from 1.0 to 6.4 mm in diameter. Airway dimensions obtained from CT data correlated with morphologic measurements (r = 0.96 for Ai and r = 0.91 for Aaw). However the CT systematically underestimated Ai and overestimated Aaw; average error (100 x (CT-morphology) / morphology) was -55% for Ai and +90% for Aaw. We used the morphology data to correct the CT measurements and reduced the average error to +23% for Ai and +7% for Aaw. This algorithm can be used to assess the structure and function of human airways.

A low cost fMRI-compatible system was developed for detecting electrodermal activity without inducing image artifact. Subject electrodermal activity was measured on the plantar surface of the foot using a standard recording circuit. Filtered analog skin conductance responses (SCR) were recorded with a general purpose, time-stamping data acquisition system. A conditioning paradigm involving painful thermal stimulation was used to demonstrate SCR detection and investigate neural correlates of conditioned autonomic activity. 128x128 pixel EPI-BOLD images were acquired with a GE 1.5T Signa scanner. Image analysis was performed using voxel-wise multiple linear regression. The covariate of interest was generated by convolving stimulus event onset with a standard hemodynamic response function. The function was time-shifted to determine optimal activation. Significance was tested using the t-statistic. Image quality was unaffected by the device, and conditioned and unconditioned SCRs were successfully detected. Conditioned SCRs correlated significantly with activity in the right anterior insular cortex. The effect was more robust when responses were scaled by SCR amplitude. The ability to measure and time register SCRs during fMRI acquisition enables studies of cognitive processes marked by autonomic activity, including those involving decision-making, pain, emotion, and addiction.

We developed 3D MR based image processing methods for biomechanical analysis of joints. These methods provide quantitative data on the morphological distribution of the joint cartilage as well as biomechanical analysis of relative translation and rotation of joints. After image data acquisition in an open MR system, the segmentation of the different joint structures was performed by a semi automatic technique based on a gray value oriented region growing algorithm. After segmentation 3D reconstructions of cartilage and bone surfaces were performed. Principal axis decomposition is used to calculate a reproducible tibia plateau based coordinate system that allows the determination of relative rotation and translation of the condyles and menisci in relation to the tibia plateau. The analysis of the femoral movement is based on a reproducible, semi automatic calculated epicondylar axis. The analysis showed a posterior translation of the meniscus and even more of the femur condyles in healthy knees and in knees with an insufficiency of the anterior cruciate ligament (ACL).

During the last several years perfusion CT techniques have been developed as an effective technique for clinically evaluating cerebral hemodynamics. Perfusion CT techniques are capable of measurings functional parameters such as tissue perfusion, blood flow, blood volume, and mean transit time and are commonly used to evaluate stroke patients. However, the quality of functional images of the brain frequently suffers from patient head motion. Because the time window for an effective treatment of stroke patient is narrow, a fast motion correction is required. The purpose of the paper is to present a fast and accurate registration technique for motion correction of multi-slice CT and to demonstrate the effects of the registration on perfusion calculation.

We present a method to construct the geometric model of the pulmonary arteries from a set of cardiac CT scan images. It is desired that the model is in rectangular meshes. The main difficulties in this work are insufficient resolution along z-direction, the requirement of the rectangular meshes, and the geometric shape of the pulmonary arteries. We present a method that is based on estimation the medial axis and the radii of the vessel along the axis. We evaluate the proposed method using a phantom data set. The proposed method can achieve good reconstructed result for the phantom data set.