Thomas Buelow

Simultaneous segmentation and tree reconstruction of the coronary arteries in MSCT images

Multislice CT angiography (MSCTA) is an emerging modality for assessing the coronary arteries. The use of MSCTA for coronary artery disease (CAD) quantification requires an assessment procedure of the coronary arteries that is automated as much as possible. We present an algorithm for the segmentation of the coronary tree with simultaneous extraction of the centerline and the tree-structure. Our approach limits the required user interaction to the placement of one landmark in the left and right main coronary artery respectively. The whole segmentation process takes about 15 s on a mid-sized PC (1GHz) including a real-time visualization of the segmentation in progress. The presented method combines a fast region expansion method (fast marching/front propagation) with heuristic reasoning. The spreading front is monitored for front-splitting enabling branch detection and simultaneous tree reconstruction of the segmented object. This approach allows for the individual treatment of tree-branches with respect to, e.g., threshold settings and reasoning on tree and sub-tree level. This approach can be applied quite generally to the segmentation of tree-like structures. The segmentation results support efficient reporting by enabling automatic generation of overview visualizations, guidance for virtual endoscopy, generation of curved MPRs along the vessels, or cross-sectional area graphs.

Biomechanically guided prone-to-supine image registration of breast MRI using an estimated reference state

The female breast undergoes large scale deformations, when the patient position is changed from the prone position where imaging is usually performed to the supine position, which is the standard surgical setting. To guide the surgical procedure, MRI data need to be aligned between these two positions. Image registration techniques which are purely intensity based usually fail, when prone and supine image data are to be aligned. To address this we use patient specific biomechanical models to provide an initial deformation of the breast prior to registration. In contrast to previous methods, we use these models to estimate the zero-gravity reference state for both the prone and supine configurations and perform the subsequent registration in this space. In this symmetric approach we incorporate non-linear material models and displacement boundary conditions on the chest wall which lead to clinically useful accuracy in the simulation and subsequent registration.

Robust segmentation of mass-lesions in contrast-enhanced dynamic breast MR images.

To develop and evaluate a computerized segmentation method for breast MRI (BMRI) mass‐lesions.

Breast deformation modelling: comparison of methods to obtain a patient specific unloaded configuration

In biomechanical simulations of the human breast, the analysed geometry is often reconstructed from in vivo medical imaging procedures. For example in dynamic contrast enhanced magnetic resonance imaging, the acquired geometry of the patients breast when lying in the prone position represents a deformed configuration that is pre-stressed by typical in vivo conditions and gravity. Thus, physically realistic simulations require consideration of this loading and, hence, establishing the undeformed configuration is an important task for accurate and reliable biomechanical modelling of the breast. We compare three different numerical approaches to recover the unloaded configuration from the loaded geometry given patient-specific biomechanical models built from prone and supine MR images. The algorithms compared are:(i) the simple inversion of gravity without the consideration of pre-stresses, (ii) an inversefinite deformation approach and (iii) afixed point type iterative approach which uses only forward simulations. It is shown that the iterative and the inverse approach produce similar zero-gravity estimates, where as the simple inversion of gravity is only appropriate for small or highly constrained deformations.

Improved sensitivity of dynamic CT with a new visualization method for radial distribution of lung nodule enhancement

For differential diagnosis of pulmonary nodules, assessment of contrast enhancement at chest CT scans after administration of contrast agent has been suggested. Likelihood of malignancy is considered very low if the contrast enhancement is lower than a certain threshold (10-20 HU). Automated average density measurement methods have been developed for that purpose. However, a certain fraction of malignant nodules does not exhibit significant enhancement when averaged over the whole nodule volume. The purpose of this paper is to test a new method for reduction of false negative results. We have investigated a method of showing not only a single averaged contrast enhancement number, but a more detailed enhancement curve for each nodule, showing the enhancement as a function of distance to boundary. A test set consisting of 11 malignant and 11 benign pulmonary lesions was used for validation, with diagnoses known from biopsy or follow-up for more than 24 months. For each nodule dynamic CT scans were available: the unenhanced native scan and scans after 60, 120, 180 and 240 seconds after onset of contrast injection (1 - 4 mm reconstructed slice thickness). The suggested method for measurement and visualization of contrast enhancement as radially resolved curves has reduced false negative results (apparently unenhancing but truly malignant nodules), and thus improved sensitivity. It proved to be a valuable tool for differential diagnosis between malignant and benign lesions using dynamic CT.

Characterization of Cystic Lesions by Spectral Mammography: Results of a Clinical Pilot Study.

ObjectivesRound lesions are a common mammographic finding, which can contribute more than 20% of overall recalls at screening. Discrimination of cystic fluid from solid tissue by spectral x-ray imaging has been demonstrated in specimen experiments. This work translates these results into a clinical pilot study to investigate the feasibility of discriminating cystic from solid lesions using spectral mammography. Materials and MethodsWomen undergoing mammography as part of their routine diagnostic workup were consented for analysis of spectral information obtained from a photon-counting mammography system. Images were analyzed retrospectively after diagnosis was confirmed with ultrasound and pathology. Well-defined solitary lesions were delineated independently by 3 expert radiologists. A breast lesion model is generated from the spectral mammography data using the energy-dependent x-ray attenuation of cyst fluid, carcinoma, and adipose and glandular tissue. From the breast lesion model, 2 spectral features are computed and combined in a 2-feature discrimination algorithm, which is evaluated in an analysis of the receiver operating characteristic curve for the task of identifying solid lesions (“positive result”). Expected outcomes on a screening population are extrapolated from this pilot study by cross-validation with bootstrapping using a 95% confidence interval (CI). ResultsThe 2-feature discrimination algorithm was evaluated on the set of 119 eligible lesions (62 solids, 57 cysts) of diameter greater than 10 mm. The area under the receiver operating characteristic curve (AUC) was 0.88 with a specificity of 61% at the 99% sensitivity level on average over all expert radiologists. Cross-validation with bootstrapping of the clinical data revealed an AUC of 0.89 (95% CI, 0.79–0.96) and a specificity of 56% (95% CI, 33%–78%) when operating the algorithm at the 99% sensitivity level. ConclusionsDiscriminating cystic from solid lesions with spectral mammography demonstrates promising results with the potential to reduce mammographic recalls. It is estimated that for each missed cancer at least 625 cystic lesions would have been correctly identified and hence would not have been needed to be recalled. Our results justify undertaking a larger reader study to refine the algorithm and determine clinically relevant thresholds to allow safe classification of cystic lesions by spectral mammography.

Filter learning and evaluation of the computer aided visualization and analysis (CAVA) paradigm for pulmonary nodules using the LIDC-IDRI database

We present a simple rendering scheme for thoracic CT datasets which yields a color coding based on local differential geometry features rather than Hounsfield densities. The local curvatures are computed on several resolution scales and mapped onto different colors, thereby enhancing nodular and tubular structures. The rendering can be used as a navigation device to quickly access points of possible chest anomalies, in particular lung nodules and lymph nodes. The underlying principle is to use the nodule enhancing overview as a possible alternative to classical CAD approaches by avoiding explicit graphical markers. For performance evaluation we have used the LIDC-IDRI lung nodule data base. Our results indicate that the nodule-enhancing overview correlates well with the projection images produced from the IDRI expert annotations, and that we can use this measure to optimize the combination of differential geometry filters.

Heterogeneity of kinetic curve parameters as indicator for the malignancy of breast lesions in DCE MRI

Dynamic contrast enhanced Breast MRI (DCE BMRI) has emerged as powerful tool in the diagnostic work-up of breast cancer. While DCE BMRI is very sensitive, specificity remains to be an issue. Consequently, there is a need for features that support the classification of enhancing lesions into benign and malignant lesions. Traditional features include the morphology and the texture of a lesion, as well as the kinetic parameters of the time-intensity curves, i.e., the temporal change of image intensity at a given location. The kinetic parameters include initial contrast uptake of a lesion and the type of the kinetic curve. The curve type is usually assigned to one of three classes: persistent enhancement (Type I), plateau (Type II), and washout (Type III). While these curve types show a correlation with the tumor type (benign or malignant), only a small sub-volume of the lesion is taken into consideration and the curve type will depend on the location of the ROI that was used to generate the kinetic curve. Furthermore, it has been shown that the curve type significantly depends on which MR scanner was used as well as on the scan parameters. Recently, it was shown that the heterogeneity of a given lesion with respect to spatial variation of the kinetic curve type is a clinically significant indicator for malignancy of a tumor. In this work we compare four quantitative measures for the degree of heterogeneity of the signal enhancement ratio in a tumor and evaluate their ability of predicting the dignity of a tumor. All features are shown to have an area under the ROC curve of between 0.63 and 0.78 (for a single feature).

Inter- and intra-observer variations in the delineation of lesions in mammograms

Many clinical and research tasks require the delineation of lesions in radiological images. There is a variety of methods available for deriving such delineations, ranging from free hand manual contouring and manual positioning of lowparameter graphical objects, to (semi-)automatic computerized segmentation methods. In this paper we investigate the impact of the chosen segmentation method on the inter-observer variability of the resulting contour. Three different methods are compared in this paper, namely (1) manual positioning of an ellipse, (2) an automatic segmentation method, coined live-segmentation, which depends on the current mouse pointer position as input information and is updated in real-time as the user hovers with the mouse over the image and (3) free form segmentation which is realized by allowing the user to pull the result of method (2) to image positions that the contour is required to pass. Each of the three methods was used by three experienced radiologists to delineate a set of 215 round breast lesion images in digital mammograms. Agreement between contours was assessed by computing the Dice coefficient. The median Dice coefficient for the ellipses placed by different readers was 0.85. The intra-reader Dice coefficient comparing ellipses and livesegmentations was 0.84, thus showing that the live-segmentation results agree with ellipse segmentations to the same extent as readers agree on the ellipse placement. Inter-observer agreement when using the live-segmentation was higher than for the ellipses (median Dice = 0.91 vs. 0.85) showing that the live-segmentation is a more reproducible alternative to the ellipse placement.

An anatomically oriented breast model for MRI

Breast cancer is the most common cancer in women in the western world. In the breast cancer care-cycle, MRIis e.g. employed in lesion characterization and therapy assessment. Reading of a single three dimensional image or comparing a multitude of such images in a time series is a time consuming task. Radiological reporting is done manually by translating the spatial position of a finding in an image to a generic representation in the form of a breast diagram, outlining quadrants or clock positions. Currently, registration algorithms are employed to aid with the reading and interpretation of longitudinal studies by providing positional correspondence. To aid with the reporting of findings, knowledge about the breast anatomy has to be introduced to translate from patient specific positions to a generic representation. In our approach we fit a geometric primitive, the semi-super-ellipsoid to patient data. Anatomical knowledge is incorporated by fixing the tip of the super-ellipsoid to the mammilla position and constraining its center-point to a reference plane defined by landmarks on the sternum. A coordinate system is then constructed by linearly scaling the fitted super-ellipsoid, defining a unique set of parameters to each point in the image volume. By fitting such a coordinate system to a different image of the same patient, positional correspondence can be generated. We have validated our method on eight pairs of baseline and follow-up scans (16 breasts) that were acquired for the assessment of neo-adjuvant chemotherapy. On average, the location predicted and the actual location of manually set landmarks are within a distance of 5.6 mm. Our proposed method allows for automatic reporting simply by uniformly dividing the super-ellipsoid around its main axis.