Lung registration using automatically detected landmarks.

OBJECTIVES: Accurate registration of lung CT images is inevitable for numerous clinical applications. Usually, nonlinear intensity-based methods are used. Their accuracy is typically evaluated using corresponding anatomical points (landmarks; e.g. bifurcations of bronchial and vessel trees) annotated by medical experts in the images to register. As image registration can be interpreted as correspondence finding problem, these corresponding landmarks can also be used in feature-based registration techniques. Recently, approaches for automated identification of such landmark correspondences in lung CT images have been presented. In this work, a novel combination of variational nonlinear intensity-based registration with an approach for automated landmark correspondence detection in lung CT pairs is presented and evaluated. METHODS: The main blocks of the proposed hybrid intensity- and feature-based registration scheme are a two-step landmark correspondence detection and the so-called CoLD (Combining Landmarks and Distance Measures) framework. The landmark correspondence identification starts with feature detection in one image followed by a blockmatching-based transfer of the features to the other image. The established correspondences are used to compute a thin-plate spline (TPS) transformation. Within CoLD, the TPS transformation is improved by minimization of an objective function consisting of a Normalized Gradient Field distance measure and a curvature regularizer; the landmark correspondences are guaranteed to be preserved by optimization on the kernel of the discretized landmark constraints. RESULTS: Based on ten publicly available end-inspiration/expiration CT scan pairs with anatomical landmark sets annotated by medical experts from the DIR-Lab database, it is shown that the hybrid registration approach is superior in terms of accuracy: The mean distance of expert landmarks is decreased from 8.46 mm before to 1.15 mm after registration, outperforming both the TPS transformation (1.68 mm) and a nonlinear registration without usage of automatically detected landmarks (2.44 mm). The improvement is statistically significant in eight of ten datasets in comparison to TPS and in nine of ten datasets in comparison to the intensity-based registration. Furthermore, CoLD globally estimates the breathing-induced lung volume change well and results in smooth and physiologically plausible motion fields of the lungs. CONCLUSIONS: We demonstrated that our novel landmark-based registration pipeline outperforms both TPS and the underlying nonlinear intensity-based registration without landmark usage. This highlights the potential of automatic landmark correspondence detection for improvement of lung CT registration accuracy.

Simulation of range imaging-based estimation of respiratory lung motion. Influence of noise, signal dimensionality and sampling patterns.

OBJECTIVES: A major problem associated with the irradiation of thoracic and abdominal tumors is respiratory motion. In clinical practice, motion compensation approaches are frequently steered by low-dimensional breathing signals (e.g., spirometry) and patient-specific correspondence models, which are used to estimate the sought internal motion given a signal measurement. Recently, the use of multidimensional signals derived from range images of the moving skin surface has been proposed to better account for complex motion patterns. In this work, a simulation study is carried out to investigate the motion estimation accuracy of such multidimensional signals and the influence of noise, the signal dimensionality, and different sampling patterns (points, lines, regions). METHODS: A diffeomorphic correspondence modeling framework is employed to relate multidimensional breathing signals derived from simulated range images to internal motion patterns represented by diffeomorphic non-linear transformations. Furthermore, an automatic approach for the selection of optimal signal combinations/patterns within this framework is presented. RESULTS: This simulation study focuses on lung motion estimation and is based on 28 4D CT data sets. The results show that the use of multidimensional signals instead of one-dimensional signals significantly improves the motion estimation accuracy, which is, however, highly affected by noise. Only small differences exist between different multidimensional sampling patterns (lines and regions). Automatically determined optimal combinations of points and lines do not lead to accuracy improvements compared to results obtained by using all points or lines. CONCLUSIONS: Our results show the potential of multidimensional breathing signals derived from range images for the model-based estimation of respiratory motion in radiation therapy.

Multivariate regression approaches for surrogate-based diffeomorphic estimation of respiratory motion in radiation therapy.

Breathing-induced location uncertainties of internal structures are still a relevant issue in the radiation therapy of thoracic and abdominal tumours. Motion compensation approaches like gating or tumour tracking are usually driven by low-dimensional breathing signals, which are acquired in real-time during the treatment. These signals are only surrogates of the internal motion of target structures and organs at risk, and, consequently, appropriate models are needed to establish correspondence between the acquired signals and the sought internal motion patterns. In this work, we present a diffeomorphic framework for correspondence modelling based on the Log-Euclidean framework and multivariate regression. Within the framework, we systematically compare standard and subspace regression approaches (principal component regression, partial least squares, canonical correlation analysis) for different types of common breathing signals (1D: spirometry, abdominal belt, diaphragm tracking; multi-dimensional: skin surface tracking). Experiments are based on 4D CT and 4D MRI data sets and cover intra- and inter-cycle as well as intra- and inter-session motion variations. Only small differences in internal motion estimation accuracy are observed between the 1D surrogates. Increasing the surrogate dimensionality, however, improved the accuracy significantly; this is shown for both 2D signals, which consist of a common 1D signal and its time derivative, and high-dimensional signals containing the motion of many skin surface points. Eventually, comparing the standard and subspace regression variants when applied to the high-dimensional breathing signals, only small differences in terms of motion estimation accuracy are found.

Automated volumes-of-interest identification for classical and atypical Parkinsonian syndrome differentiation using T2' MR imaging.

OBJECTIVES: In clinical routine, patients with classical Parkinsonian syndromes (CPS) need to be differentiated from those with atypical Parkinsonian syndromes (APS), particularly with respect to prognosis and treatment decision. To date, this diagnosis is mainly based on clinical criteria, leading to failure rates up to 25%, motivating the development of image-based decision support systems. Magnetic resonance imaging (MRI) and in particular T2´ image sequences have been suggested as a potential marker for differential diagnosis. The aim of this study was to investigate whether automatically identified T2´ volumes-of-interest (VOIs) can be used for an automatic differentiation of CPS and APS patients. MATERIAL AND METHODS: 74 MRI datasets were available for this hypothesis generating trial, including image sequences from 24 healthy subjects, 33 CPS and 17 APS patients. First, a problem-specific reference atlas was generated using the healthy control datasets. Next, patients' datasets were registered to the atlas. Voxel-wise t-tests, reflecting significance levels of T2´ value differences between CPS and APS patients, were then applied for calculation of a p-map. Finally, the calculated p-map was thresholded and a connected component analysis was performed for final VOI detection. In parallel, manually defined VOIs were determined in grey and white matter for comparison. RESULTS: Three VOIs in parts of the basal ganglia and the left occipital lobe were automatically identified by the presented method. There was a trend for higher area under the curve on multivariable receiver operating characteristic curves for automatically determined VOIs over manually defined VOIs (0.939 vs. 0.818, p = 0.0572). CONCLUSION: The diagnostic role of automatically defined VOIs in differentiation of CPS and APS patients based on T2´ image sequences should be further investigated.

Comparison of 10 TTP and Tmax estimation techniques for MR perfusion-diffusion mismatch quantification in acute stroke.

BACKGROUND AND PURPOSE: The mismatch between lesions identified in perfusion- and diffusion-weighted MR imaging is typically used to identify tissue at risk of infarction in acute stroke. The purpose of this study was to analyze the variability of mismatch volumes resulting from different time-to-peak or time-to-maximum estimation techniques used for hypoperfused tissue definition. MATERIALS AND METHODS: Data of 50 patients with middle cerebral artery stroke and intracranial vessel occlusion imaged within 6 hours of symptom onset were analyzed. Therefore, 10 different TTP/Tmax techniques and delay thresholds between +2 and +12 seconds were used for calculation of perfusion lesions. Diffusion lesions were semiautomatically segmented and used for mismatch quantification after registration. RESULTS: Mean volumetric differences up to 40 and 100 mL in individual patients were found between the mismatch volumes calculated by the 10 TTP/Tmax estimation techniques for typically used delay thresholds. The application of typical criteria for the identification of patients with a clinically relevant mismatch volume resulted in different mismatch classifications in ≤24% of all cases, depending on the TTP/Tmax estimation method used. CONCLUSIONS: High variations of tissue-at-risk volumes have to be expected when using different TTP/Tmax estimation techniques. An adaption of different techniques by using correction formulas may enable more comparable study results until a standard has been established by agreement.

Intranidal signal distribution in post-contrast time-of-flight MRA is associated with rupture risk factors in arteriovenous malformations.

PURPOSE: To evaluate if arteriovenous malformations (AVMs) that are associated with a high rupture risk (HRR) are represented by different intranidal Time-of-Flight (TOF) magnetic resonance angiography intensity distributions compared to those with presumably low rupture risk (LRR). METHODS: Fifty post-contrast TOF datasets of patients with an AVM were analyzed in this study. The patients were classified to the HRR group in case of a deep location, presence of exclusive deep venous drainage, previous hemorrhagic event or a combination thereof. For each TOF dataset, the AVM nidus was semi-automatically delineated and used for histogram extraction. Each histogram was analyzed by calculating the skewness, kurtosis, mean and median intensity and full-width-half-maximum. Statistical analysis was performed using parameter-wise two-sided t-tests of the parameters between the two groups. RESULTS: Based on morphological analysis, 21 patients were classified to the HRR and 29 patients to the LRR group. Statistical analysis revealed that TOF intensity distributions of HRR AVMs exhibit a significant higher skewness (p=0.0005) parameter compared to LRR AVMs. Contrary to these findings, no significant differences were found for the other parameters evaluated. CONCLUSION: Intranidal flow heterogeneity, for example, caused by turbulent flow conditions, may play an important role for risk of a hemorrhage. An analysis of post-contrast TOF intensities within the nidus of an AVM may offer simple and valuable information for clinical risk estimation of AVMs and needs to be tested prospectively.

Image analysis and modeling in medical image computing. Recent developments and advances.

BACKGROUND: Medical image computing is of growing importance in medical diagnostics and image-guided therapy. Nowadays, image analysis systems integrating advanced image computing methods are used in practice e.g. to extract quantitative image parameters or to support the surgeon during a navigated intervention. However, the grade of automation, accuracy, reproducibility and robustness of medical image computing methods has to be increased to meet the requirements in clinical routine. OBJECTIVES: In the focus theme, recent developments and advances in the field of modeling and model-based image analysis are described. The introduction of models in the image analysis process enables improvements of image analysis algorithms in terms of automation, accuracy, reproducibility and robustness. Furthermore, model-based image computing techniques open up new perspectives for prediction of organ changes and risk analysis of patients. METHODS: Selected contributions are assembled to present latest advances in the field. The authors were invited to present their recent work and results based on their outstanding contributions to the Conference on Medical Image Computing BVM 2011 held at the University of Lübeck, Germany. All manuscripts had to pass a comprehensive peer review. RESULTS: Modeling approaches and model-based image analysis methods showing new trends and perspectives in model-based medical image computing are described. Complex models are used in different medical applications and medical images like radiographic images, dual-energy CT images, MR images, diffusion tensor images as well as microscopic images are analyzed. The applications emphasize the high potential and the wide application range of these methods. CONCLUSIONS: The use of model-based image analysis methods can improve segmentation quality as well as the accuracy and reproducibility of quantitative image analysis. Furthermore, image-based models enable new insights and can lead to a deeper understanding of complex dynamic mechanisms in the human body. Hence, model-based image computing methods are important tools to improve medical diagnostics and patient treatment in future.

Automatic correction of gaps in cerebrovascular segmentations extracted from 3D time-of-flight MRA datasets.

OBJECTIVES: Exact cerebrovascular segmentations are required for several applications in today's clinical routine. A major drawback of typical automatic segmentation methods is the occurrence of gaps within the segmentation. These gaps are typically located at small vessel structures exhibiting low intensities. Manual correction is very time-consuming and not suitable in clinical practice. This work presents a post-processing method for the automatic detection and closing of gaps in cerebrovascular segmentations. METHODS: In this approach, the 3D centerline is calculated from an available vessel segmentation, which enables the detection of corresponding vessel endpoints. These endpoints are then used to detect possible connections to other 3D centerline voxels with a graph-based approach. After consistency check, reasonable detected paths are expanded to the vessel boundaries using a level set approach and combined with the initial segmentation. RESULTS: For evaluation purposes, 100 gaps were artificially inserted at non-branching vessels and bifurcations in manual cerebrovascular segmentations derived from ten Time-of-Flight magnetic resonance angiography datasets. The results show that the presented method is capable of detecting 82% of the non-branching vessel gaps and 84% of the bifurcation gaps. The level set segmentation expands the detected connections with 0.42 mm accuracy compared to the initial segmentations. A further evaluation based on 10 real automatic segmentations from the same datasets shows that the proposed method detects 35 additional connections in average per dataset, whereas 92.7% were rated as correct by a medical expert. CONCLUSION: The presented approach can considerably improve the accuracy of cerebrovascular segmentations and of following analysis outcomes.

Analysis of the influence of 4D MR angiography temporal resolution on time-to-peak estimation error for different cerebral vessel structures.

BACKGROUND AND PURPOSE: Time-resolved MRA imaging is a promising technique for blood flow evaluation in case of cerebrovascular malformations. Unfortunately, 4D MRA imaging is a trade-off between spatial and temporal resolution. The aim of this study was to investigate the influence of temporal resolution on the error associated with TTP estimation from indicator dilution curves derived from different vascular structures. MATERIALS AND METHODS: Monte Carlo simulation was performed to compute indicator dilution curves with known criterion standard TTP at temporal resolutions between 0.1 and 5 seconds. TTPs were estimated directly and by using 4 hemodynamic models for each curve and were compared with criterion standard TTP. Furthermore, clinical evaluation was performed by using 226 indicator dilution curves from different vessel structures obtained from clinical datasets. The temporal resolution was artificially decreased, and TTPs were estimated and compared with those obtained at the original temporal resolutions. The results of the clinical evaluations were further stratified for different vessel structures. RESULTS: The results of both evaluations show that the TTP estimation error increases exponentially when one lowers the temporal resolution. TTP estimation by using hemodynamic model curves leads to lower estimation errors compared with direct estimation. A temporal resolution of 1.5 seconds for arteries and 2.5 seconds for venous and arteriovenous malformation vessel structures appears to be reasonable to achieve TTP estimations adequate for clinical application. CONCLUSIONS: Different vessel structures require different temporal resolutions to enable comparable TTP estimation errors, which should be considered for achieving a case-optimal temporal and spatial resolution.

Fuzzy-based vascular structure enhancement in Time-of-Flight MRA images for improved segmentation.

OBJECTIVES: Cerebral vascular malformations might lead to strokes due to occurrence of ruptures. The rupture risk is highly related to the individual vascular anatomy. The 3D Time-of-Flight (TOF) MRA technique is a commonly used non-invasive imaging technique for exploration of the vascular anatomy. Several clinical applications require exact cerebrovascular segmentations from this image sequence. For this purpose, intensity-based segmentation approaches are widely used. Since small low-contrast vessels are often not detected, vesselness filter-based segmentation schemes have been proposed, which contrariwise have problems detecting malformed vessels. In this paper, a fuzzy logic-based method for fusion of intensity and vesselness information is presented, allowing an improved segmentation of malformed and small vessels at preservation of advantages of both approaches. METHODS: After preprocessing of a TOF dataset, the corresponding vesselness image is computed. The role of the fuzzy logic is to voxel-wisely fuse the intensity information from the TOF dataset with the corresponding vesselness information based on an analytically designed rule base. The resulting fuzzy parameter image can then be used for improved cerebrovascular segmentation. RESULTS: Six datasets, manually segmented by medical experts, were used for evaluation. Based on TOF, vesselness and fused fuzzy parameter images, the vessels of each patient were segmented using optimal thresholds computed by maximizing the agreement to manual segmentations using the Tanimoto coefficient. The results showed an overall improvement of 0.054 (fuzzy vs. TOF) and 0.079 (fuzzy vs. vesselness). Furthermore, the evaluation has shown that the method proposed yields better results than statistical Bayes classification. CONCLUSION: The proposed method can automatically fuse the benefits of intensity and vesselness information and can improve the results of following cerebrovascular segmentations.

Automatic brain segmentation in Time-of-Flight MRA images.

OBJECTIVES: Cerebral vascular malformations might, caused by ruptures, lead to strokes. The rupture risk depends to a great extent on the individual anatomy of the vasculature. The 3D Time-of-Flight (TOF) MRA technique is one of the most commonly used non-invasive imaging techniques to obtain knowledge about the individual vascular anatomy. Unfortunately TOF images exhibit drawbacks for segmentation and direct volume visualization of the vasculature. To overcome these drawbacks an initial segmentation of the brain tissue is required. METHODS: After preprocessing of the data is applied the low-intensity tissues surrounding the brain are segmented using region growing. In a following step this segmentation is used to extract supporting points at the border of the brain for a graph-based contour extraction. Finally a consistency check is performed to identify local outliers which are corrected using non-linear registration. RESULTS: A quantitative validation of the method proposed was performed on 18 clinical datasets based on manual segmentations. A mean Dice coefficient of 0.989 was achieved while in average 99.56% of all vessel voxels were included by the brain segmentation. A comparison to the results yielded by three commonly used tools for brain segmentation revealed that the method described achieves better results, using TOF images as input, which are within the inter-observer variability. CONCLUSION: The method suggested allows a robust and automatic segmentation of brain tissue in TOF images. It is especially helpful to improve the automatic segmentation or direct volume rendering of the cerebral vascular system.

Advances in medical image computing.

OBJECTIVES: Medical image computing has become a key technology in high-tech applications in medicine and an ubiquitous part of modern imaging systems and the related processes of clinical diagnosis and intervention. Over the past years significant progress has been made in the field, both on methodological and on application level. Despite this progress there are still big challenges to meet in order to establish image processing routinely in health care. In this issue, selected contributions of the German Conference on Medical Image Processing (BVM) are assembled to present latest advances in the field of medical image computing. METHODS: The winners of scientific awards of the German Conference on Medical Image Processing (BVM) 2008 were invited to submit a manuscript on their latest developments and results for possible publication in Methods of Information in Medicine. Finally, seven excellent papers were selected to describe important aspects of recent advances in the field of medical image processing. RESULTS: The selected papers give an impression of the breadth and heterogeneity of new developments. New methods for improved image segmentation, non-linear image registration and modeling of organs are presented together with applications of image analysis methods in different medical disciplines. Furthermore, state-of-the-art tools and techniques to support the development and evaluation of medical image processing systems in practice are described. CONCLUSIONS: The selected articles describe different aspects of the intense development in medical image computing. The image processing methods presented enable new insights into the patient's image data and have the future potential to improve medical diagnostics and patient treatment.

Integrated segmentation and non-linear registration for organ segmentation and motion field estimation in 4D CT data.

OBJECTIVES: The development of spatiotemporal tomographic imaging techniques allows the application of novel techniques for diagnosis and therapy in the medical routine. However, in consequence to the increasing amount of image data automatic methods for segmentation and motion estimation are required. In adaptive radiation therapy, registration techniques are used for the estimation of respiration-induced motion of pre-segmented organs. In this paper, a variational approach for the simultaneous computation of segmentations and a dense non-linear registration of the 3D images of the sequence is presented. METHODS: In the presented approach, a variational region-based level set segmentation of the structures of interest is combined with a diffusive registration of the spatial images of the sequence. We integrate both parts by defining a new energy term, which allows us to incorporate mutual prior information in order to improve the segmentation as well as the registration quality. RESULTS: The presented approach was utilized for the segmentation of the liver and the simultaneous estimation of its respiration-induced motion based on four-dimensional thoracic CT images. For the considered patients, we were able to improve the results of the segmentation and the motion estimation, compared to the conventional uncoupled methods. CONCLUSIONS: Applied in the field of radiation therapy of thoracic tumors, the presented integrated approach turns out to be useful for simultaneous segmentation and registration by improving the results compared to the application of the methods independently.

3D segmentation of the left ventricle combining long- and short-axis MR images.

OBJECTIVES: Segmentation of the left ventricle (LV) is required to quantify LV remodeling after myocardial infarction. Therefore spatiotemporal cine MR sequences including long-axis and short-axis images are acquired. In this paper a new segmentation method for fast and robust segmentation of the left ventricle is presented. METHODS: The new approach considers the position of the mitral valve and the apex as well as the long-axis contours to generate a 3D LV surface model. The segmentation result can be checked and adjusted in the short-axis images. Finally quantitative parameters were extracted. RESULTS: For evaluation the LV was segmented in eight datasets of the same subject by two medical experts using a contour drawing tool and the new segmentation tool. The results of both methods were compared concerning interaction time and intra- and inter-observer variance. The presented segmentation method proved to be fast. The mean difference and standard deviation of all parameters are decreased. In case of intra-observer comparison e.g. the mean ESV difference is reduced from 8.8% to 0.5%. CONCLUSION: A semi-automatic LV segmentation method has been developed that combines long- and short-axis views. Using the presented approach the intra- and interobserver difference as well as the time for the segmentation process are decreased. So the semi-automatic segmentation using long- and short-axis information proved to be fast and robust for the quantification of LV mass and volume properties.

Computation of a probabilistic statistical shape model in a maximum-a-posteriori framework.

OBJECTIVES: When analyzing shapes and shape variabilities, the first step is bringing those shapes into correspondence. This is a fundamental problem even when solved by manually determining exact correspondences such as landmarks. We developed a method to represent a mean shape and a variability model for a training data set based on probabilistic correspondence computed between the observations. METHODS: First, the observations are matched on each other with an affine transformation found by the Expectation-Maximization Iterative-Closest-Points (EM-ICP) registration. We then propose a maximum-a-posteriori (MAP) framework in order to compute the statistical shape model (SSM) parameters which result in an optimal adaptation of the model to the observations. The optimization of the MAP explanation is realized with respect to the observation parameters and the generative model parameters in a global criterion and leads to very efficient and closed-form solutions for (almost) all parameters. RESULTS: We compared our probabilistic SSM to a SSM based on one-to-one correspondences and the PCA (classical SSM). Experiments on synthetic data served to test the performances on non-convex shapes (15 training shapes) which have proved difficult in terms of proper correspondence determination. We then computed the SSMs for real putamen data (21 training shapes). The evaluation was done by measuring the generalization ability as well as the specificity of both SSMs and showed that especially shape detail differences are better modeled by the probabilistic SSM (Hausdorff distance in generalization ability Re approximately 25% smaller). CONCLUSIONS: The experimental outcome shows the efficiency and advantages of the new approach as the probabilistic SSM performs better in modeling shape details and differences.

Virtual reality simulator for the training of lumbar punctures.

OBJECTIVES: Lumbar puncture (LP) is performed by inserting a needle into the spinal canal to extract cerebrospinal fluid for diagnostic purposes. A virtual reality (VR) lumbar puncture simulator based on real patient data has been developed and evaluated. METHODS: A haptic device with six degrees of freedom is used to steer the virtual needle and to generate feedback forces that resist needle insertion and rotation. An extended haptic volume-rendering approach is applied to calculate forces. This approach combines information from segmented data and original CT data which contributes density information in unsegmented image structures. The system has been evaluated in a pilot study with medical students. Participants of two groups, a training and a control group, completed different first training protocols. User performance has been recorded during a second training session to measure the training effect. Furthermore user acceptance has been evaluated in a questionnaire using a 6-point Likert scale with eight items. RESULTS: Forty-two medical students in two groups evaluated the system. Trained users performed better than less trained users (an average of 39% successfully completed virtual LPs compared to 30%). Findings of the questionnaire show that the simulator is very well accepted. E.g. the users agree that training with such a simulator is useful (Likert grade of 1.5 +/- 0.7 with 1 = "strongly agree" and 6 = "strongly disagree"). CONCLUSIONS: Results show that the VR LP simulator gives a realistic haptic and visual impression of the needle insertion and enables new insights into the anatomy of the lumbar region. It offers a new way for increasing skills of students and young residents before applying an LP in patients.

Medical image computing for computer-supported diagnostics and therapy. Advances and perspectives.

OBJECTIVES: Medical image computing has become one of the most challenging fields in medical informatics. In image-based diagnostics of the future software assistance will become more and more important, and image analysis systems integrating advanced image computing methods are needed to extract quantitative image parameters to characterize the state and changes of image structures of interest (e.g. tumors, organs, vessels, bones etc.) in a reproducible and objective way. Furthermore, in the field of software-assisted and navigated surgery medical image computing methods play a key role and have opened up new perspectives for patient treatment. However, further developments are needed to increase the grade of automation, accuracy, reproducibility and robustness. Moreover, the systems developed have to be integrated into the clinical workflow. METHODS: For the development of advanced image computing systems methods of different scientific fields have to be adapted and used in combination. The principal methodologies in medical image computing are the following: image segmentation, image registration, image analysis for quantification and computer assisted image interpretation, modeling and simulation as well as visualization and virtual reality. Especially, model-based image computing techniques open up new perspectives for prediction of organ changes and risk analysis of patients and will gain importance in diagnostic and therapy of the future. RESULTS: From a methodical point of view the authors identify the following future trends and perspectives in medical image computing: development of optimized application-specific systems and integration into the clinical workflow, enhanced computational models for image analysis and virtual reality training systems, integration of different image computing methods, further integration of multimodal image data and biosignals and advanced methods for 4D medical image computing. CONCLUSIONS: The development of image analysis systems for diagnostic support or operation planning is a complex interdisciplinary process. Image computing methods enable new insights into the patient's image data and have the future potential to improve medical diagnostics and patient treatment.

Territorial and microvascular perfusion impairment in brain arteriovenous malformations.

BACKGROUND AND PURPOSE: Both the existence and clinical relevance of a steal phenomenon in brain arteriovenous malformations (AVMs) remains a matter of debate. This study aimed to assess perfusion in the brain adjacent to brain AVMs and to relate these to macrovascular blood flow in a single measurement. MATERIALS AND METHODS: Twenty consecutive patients with AVMs with a median age of 37 years were evaluated by 3T MR imaging by using 3D time-resolved MR angiography to determine blood flow and perfusion patterns. Cerebral perfusion was estimated by using an arterial spin-labeling technique in vascular territories around the nidus and in symmetric regions of interest in the ipsilateral and contralateral hemispheres. Mapping of concentric shells around the nidus was used to define the immediate and adjacent brain and relative perfusion reductions >20% of baseline, termed perinidal dip (PND). RESULTS: A significant reduction in perfusion ratios between ipsilateral and contralateral hemispheres remote to the AVMs was demonstrated in the middle and posterior cerebral artery territories. PND was detected in 5 patients, and 17 patients overall showed reduced perfusion in the perinidal region on visual inspection. There was a negative correlation of the hemispheric territorial perfusion with the affected/nonaffected inflow time ratio (R = -0.402, P = .015). CONCLUSIONS: The perfusion impairment in vascular territories adjacent to brain AVMs that we identified as PND may reflect the existence of 2 levels of perfusion impairment: a territorial and a microvascular perfusion disturbance. Although the hemispheric asymmetry in territorial perfusion seems the result of arterioarterial redistribution, the PND was inhomogeneously distributed within a single vascular territory and thus might result from low perfusion pressure in small arteries and arterioles.

Motion artifact reducing reconstruction of 4D CT image data for the analysis of respiratory dynamics.

OBJECTIVES: Respiratory motion represents a major problem in radiotherapy of thoracic and abdominal tumors. Methods for compensation require comprehensive knowledge of underlying dynamics. Therefore, 4D (= 3D + t) CT data can be helpful. But modern CT scanners cannot scan a large region of interest simultaneously. So patients have to be scanned in segments. Commonly used approaches for reconstructing the data segments into 4D CT images cause motion artifacts. In order to reduce the artifacts, a new method for 4D CT reconstruction is presented. The resulting data sets are used to analyze respiratory motion. METHODS: Spatiotemporal CT image sequences of lung cancer patients were acquired using a multi-slice CT in cine mode during free breathing. 4D CT reconstruction was done by optical flow based temporal interpolation. The resulting 4D image data were compared with data generated by the commonly used nearest neighbor reconstruction. Subsequent motion analysis is mainly concerned with tumor mobility. RESULTS: The presented optical flow-based method enables the reconstruction of 3D CT images at arbitrarily chosen points of the patient's breathing cycle. A considerable reduction of motion artifacts has been proven in eight patient data sets. Motion analysis showed that tumor mobility differs strongly between the patients. CONCLUSIONS: Due to the proved reduction of motion artifacts, the optical flow-based 4D CT reconstruction offers the possibility of high-quality motion analysis. Because the method is based on an interpolation scheme, it additionally has the potential to enable the reconstruction of 4D CT data from a lesser number of scans.