· · · Repository hosted by TU Delft Library

Digital breast tomosynthesis (DBT) aims for improving the diagnosis of breast cancer and reducing the false positive rates by going from 2D projection mammography to 3D volume information. With the acquisition of a series of projection images, taken over a limited angular range, DBT allows for tomographic reconstruction with high in-plane but reduced depth resolution. Therefore, anatomical structures get blurred along the depth direction and produce out-of-plane artifacts. Prominent streak artifacts can be observed for high-contrast objects such as calcifications, which degrade the image quality, see for example Figure 8(a) - (f) in [4]. In this work, a second pass method for correcting these streak-like artifacts is introduced. For evaluation of this method, a software based breast phantom has been developed from segmented MRI breast data.

Diagnosis and treatment decisions of cerebrovascular diseases are currently based on structural information like the endovascular lumen. In future, clinical diagnosis will increasingly be based on functional information which gives direct information about the physiological parameters and, hence, is a direct measure for the severity of the pathology. In this context, an important functional quantity is the volumetric blood flow over time. The proposed flow quantification method uses contrasted X-ray images from cerebrovascular interventions and a model of contrast agent dispersion to estimate the flow parameters from the spatial and temporal development of the contrast agent concentration through the vascular system. To evaluate the model-based blood flow quantification under realistic circumstances, dedicated cerebrovascular data has been acquired during clinical interventions. To this aim, a clinical protocol for this novel procedure has been defined and optimized. For the verification of the measured flow results ultrasound Doppler measurements have been performed acting as ground truth. The clinical data available so far indicates the ability of the proposed flow model to explain the in-vivo transport of contrast agent in blood. The flow quantification results show good correspondence of flow waveform and mean volumetric flow rate with the accomplished ultrasound measurements before angiography. Hence, the method has the potential to give sufficiently accurate, quantitative flow estimates for clinical practice and to be an alternative to ultrasound Doppler.

Rational: An experiment is described whose purpose is to extract those image characteristics that radiologists use in diagnostic decision making. It is one phase in the development of a clinical decisionsupport system to assist radiologists using DCE-MRI (dynamic contrast-enhanced MRI) for breast cancer diagnosis. The results of this experiment will be used to develop a case-based reasoning system, whichrelies on presenting prior similar cases with known diagnosis froma database to aid decision making. Methods: Clinical similarity formass lesions was established by four expert radiologists who systematically sorted lesions visualized by DCE-MRI into similarity clusters using a proprietary software tool. Cognitive analysis was used toidentify the relevant perceptual features characterizing each cluster, such that the list of features and clusters define the clinicalsimilarity. There were no constraints on the number or size of clusters that could be created. The radiologists first individually clustered a total of 214 lesions. A subsequent phase required all radiologists to agree on both a cluster designation and assignment of eachlesion into a cluster. Results: Radiologists created individually10, 10, 12, and 16 clusters. Of this initial cluster assignment, there was unanimous agreement in ~20% of the lesions, and majority agreement of ~60%. The final consensus assignment created 16 clusters; two consisted of all malignant lesions; two consisted of a majority of benign lesions, three were large approximately equal mixes of benign and malignant lesions, while the remaining nine were small clusters representing lesions with clinically relevant special characteristics of low clinical prevalence. Conclusions: The cognitive analysis revealed that the image characteristics differentiating the clusters are highly correlated to BI-RADS lesion descriptors. The radiologists were excellent in clustering certain malignant lesions, very good with some benign lesions; while as expected there was large variability in the majority of lesions.

In a Luminescent Solar Concentrator (LSC), short-wavelength light isconverted by a luminescent material into long-wavelength light, which is guided towards a photovoltaic cell. In principle, an LSC allows for high concentration, but in practice this is prevented by lossmechanisms like limited sunlight absorption, limited quantum efficiency and high self absorption. To tackle these problems, a suitable luminescent material is needed. Another important loss mechanism is the escape of luminescent radiation into directions that do not stayinside the light guide. To reduce this amount, wavelength-selectivefilters can be applied that reflect the luminescent radiation back into the light guide while transmitting the incident sunlight. In this paper, we discuss experiments and simulations of new luminescent and filter materials. We will introduce a phosphor with close-to-optimal luminescent properties. A problem for use in an LSC is the largescattering of this material; we will discuss possible solutions forthis. Furthermore, we will discuss the use of broad-band cholesteric filters in combination with this phosphor.

Electromagnetic (EM) tracking has been recognized as a valuable toolfor tracking the interventional devices in procedures such as lungand liver biopsy and ablation. The advantage of this technology overconventional X-ray fluoroscopy or CT-guided procedures is its real-time connection to the 3D volumetric roadmap of a patient’s anatomywhile the intervention is performed. EM-based guidance requires tracking of the tip of the interventional device, transforming the location of the device with pre-operative CT images, and superimposing the device in the 3D images to assist physician to complete the procedure more effectively. A key requirement of this integration of datais to find automatically the mapping between EM and CT coordinate systems. Thus, skin fiducial sensors are attached to patients before acquiring the pre-operative CTs. Then, those sensors can be recognized in both CT and EM coordinate systems to calculate the transformation matrix. In order to automate the EM-based navigation workflow andreduce procedural preparation time, an automatic fiducial detectionmethod is proposed to obtain the centroids of the sensors from thepre-operative CT. The approach has been applied to 13 rabbit datasets derived from an animal study, and numerical results show that it is a reliable and efficient method for use in EM-guided application.

Computed Tomography (CT) has been widely used for assisting lung cancer detection/diagnosis and treatment. In lung cancer diagnosis, suspect lesions or regions of interest (ROIs) are usually analyzed in screening CT scans, and CT-based image-guided minimally invasive procedures are performed for further diagnosis through bronchoscopic orpercutaneous approaches. Thus, ROI segmentation is a preliminary butvital step for abnormality detection, procedural planning, and intra-procedural guidance. In lung cancer diagnosis, such ROIs can be tumors, lymph nodes, nodules, etc. They may vary in size, shape, and other complication phenomena. Manual segmentation approaches are timeconsuming, user-biased, and cannot guarantee reproducible results.Automatic methods do not require user input, but they are usually highly application-dependent. To counterbalance among efficiency, accuracy, and robustness, considerable efforts have been contributed tosemi-automatic strategies, which enable full user control, while minimizing human interactions. Among available semi-automatic approaches, the live-wire algorithm has been recognized as a valuable tool for segmentation of a wide range of ROIs from chest CT images. In thispaper, the traditional 2D live-wire method is revisited and improved for 3D ROI segmentation. In the experiments, the proposed approachis applied to a set of anatomical ROIs from 3D chest CT images, andthe results are compared with the segmentation derived from previous evaluated live-wire-based approaches.

Filtered back-projection (FBP) has been commonly used as an efficient and robust reconstruction technique in tomographic X-ray imagingduring the last decades. For limited angle tomography acquisitions such as digital breast tomosynthesis, however, standard FBP reconstruction algorithms provide poor results and give rise to image artifacts due to the limited angular range and the coarse angular sampling.Therefore, iterative algorithms are often used in digital breast tomosynthesis since they potentially yield a reconstructed image thatis in better accordance with the measured data. In this work, a generalized FBP algorithm is presented, which uses the filtered projection data of all acquired views for back-projection along one direction in order to compute an image that is similar to an iteratively calculated one. The proposed method yields a computationally efficientgeneralized FBP algorithm for digital breast tomosynthesis, which provides similar image quality as iterative reconstruction techniqueswhile preserving the ability for region of interest reconstructions. Both a small number of views and a limited angular range can be handled with the generalized FBP while common FBP reconstruction yields a severe loss of the average value, which will be demonstrated onsimulated breast tomosynthesis data. Moreover, due to the filter computation as the pseudo-inverse operator, the reconstructed image provides an optimal solution in the least-squares sense, which minimizes the error between the measured data and the reprojections of thereconstructed image.

CT-fluoroscopy (CTF) is an efficient imaging method for guiding percutaneous lung interventions. During CTF-guided biopsy procedure, three to ten axial sectional images are captures in a very short time period to provide nearly real-time feedback to advise adjustment of the needle as it is advanced towards the target lesion. However, thisprocedure may require frequent scans and cause unnecessary radiation exposure to physicians and technicians. Its response to respiratory movements is limited and only provides narrow local anatomical dynamics. To better utilize CTF guidance, we propose a fast CT-CTF registration algorithm with respiratory motion estimation for image-guided lung intervention using electromagnetic (EM) guidance. With the pre-procedural exhale and inhale CT scans, it would be possible to estimate a series of CT images of the same patient at different respiratory phases. Then, once CTF images are captured during the intervention, our algorithm can choose the best respiratory phase-matched 3DCT image and performs a fast deformable registration to warp the 3DCT toward CTF. The new 3D CT image can then be used by the interventional system . Compared to the traditional repetitive CTF guidance,the registered CT integrates both 3D volumetric patient data and local nearly real-time anatomy for more effective and efficient guidance. Therefore, CTF is used as a nearly real-time sensor to overcomethe discrepancies between static pre-procedural CT and the patient’sanatomy, so as to provide global guidance that may be supplementedwith electromagnetic (EM) tracking and reduce the number of CTF scans. The comparative results using simulated and real data showed thatour fast CT-CTF algorithm can achieve better registration accuracythan using traditional 3D algorithms for CT-CTF registration.

The 3D fusion of tracked ultrasound with a diagnostic CT image has multiple benefits in a variety of interventional applications for oncology. Still, manual registration is a considerable drawback to theclinical workflow and hinders the widespread clinical adoption of this technique. In this paper, we propose a method to allow for an image-based automated registration, aligning multimodal images of the liver. We adopt a model-based approach that rigidly matches segmentedliver shapes from ultrasound (U/S) and diagnostic CT imaging. Towards this end, a novel method which combines a dynamic region-growingmethod with a graph-based segmentation framework is introduced to address the challenging problem of liver segmentation from U/S. The method is able to extract liver boundary from U/S images after a partial surface is generated near the principal vector from an electromagnetically tracked U/S liver sweep. The liver boundary is subsequently expanded by modeling the problem as a graph-cut minimization scheme, where cost functions used to detect optimal surface topology aredetermined from adaptive priors of neighboring surface points. Thisallows including boundaries affected by shadow areas by compensatingfor varying levels of contrast. The segmentation of the liver surface is performed in 3D space for increased accuracy and robustness. The method was evaluated in a study involving 8 patients undergoing biopsy or radiofrequency ablation of the liver, yielding promising surface segmentation results based on ground-truth comparison. The proposed extended segmentation technique improved the fiducial landmarkregistration error compared to a point-based registration (7.2mm vs. 10.2mm on average, respectively), while yielding a statistically insignificant differences in tumor target registration error (p > 0.05) compared to state-of-the-art methods.

Recently introduced combined PET/MR scanners need to handle the specific problem that a limited MR field of view sometimes truncates armor body contours, which prevents an accurate calculation of PET attenuation correction maps. Such maps of attenuation coefficients overbody structures are required for a quantitatively correct PET imagereconstruction. This paper addresses this problem by presenting a method that segments a preliminary reconstruction type of PET images,time of flight non-attenuation corrected (ToF-NAC) images, and outlining a processing pipeline that compensates the arm or body truncation with this segmentation. The impact of this truncation compensation is demonstrated together with a comparison of two segmentation methods, simple gray value threshold segmentation and a watershed algorithm on a gradient image. Our results indicate that with truncationcompensation a clinically tolerable quantitative SUV error is robustly achievable.

Currently, photoplethysmograms (PPGs) are mostly used to determine apatient's blood oxygenation and pulse rate. However, PPG morphologyconveys more information about the patient's cardiovascular status.Extracting this information requires measuring clean PPG waveformsthat are free of artifacts. PPGs are highly susceptible to motion, which can distort the PPG derived data. Part of the motion artifactsare considered to result from sensor-tissue motion and sensor deformation. It is hypothesized that these motion artifacts correlate withmovement of the sensor with respect to the skin. This hypothesis has been proven true in a laboratory setup. In-vitro PPGs have been measured in a skin perfusion phantom that is illuminated by a laser diode. Optical motion artifacts are generated in the PPG by translating the laser diode with respect to the PPG photodiode. The optical motion artifacts have been reduced significantly in-vitro, by using anormalized least-mean-square algorithm with only a single coefficient that uses the laser's displacement as a reference for the motion artifacts. Laser displacement has been measured accurately via self-mixing interferometry by a compact laser diode with a ball lens integrated into the package, which can be easily integrated into a commercial sensor.

The proposed method simultaneously reconstructs activity and attenuation distribution of SPECT scans without usage of additional transmission scans. Moreover, in contrast to other approaches, it effectively prevents cross-talk artefacts by using a-priori atlas data and by labelling each organ with homogeneous attenuation values. The method generates a 3D-shape model of the patient and, in order to improve overall consistency between measured and estimated SPECT sinogram, modifications to the activity- and attenuation estimation are performed iteratively. Several reconstructions of patient and simulated SPECT data were investigated and reliable convergence behaviour as well as good agreement with reference images could be observed.

Temporal subtraction techniques using 2D image registration improve the detectability of interval changes from chest radiographs. Although such methods are well known for some time they are not widely used in radiologic practice. The reason are strong pose differences between these follow-up acquisitions with a time interval of months to years in between. Such strong perspective differences occur in a reasonable number of cases. They cannot be compensated by available image registration methods and thus mask interval changes to be undetectable. A method is proposed to estimate a 3D pose difference by the adaptation of a 3D rib cage model to both projections. The difference between both is then compensated for, thus producing a subtraction image with virtually no change in pose. No 3D image data is used. The accuracy of pose estimation is validated with chest phantom images under controlled geometric conditions.

A new method using a thin-film multilayer filter is described to couple light from high-power LEDs into a thin light guide such as an LCD backlight. Light emitted below the critical angle is reflected back to the LED and recycled. Large-angle emitted light passes the filter and is transported by total internal reflection in the light guide. The light guide can be as thin as 0.3mm for an LED of 1x1mm2, and the best coupling efficiency is estimated to be 82%. With this approach, a backlight system can be greatly simplified but also compactcollimators can be realized. In this paper the optical design and testing of the filter is described, and a 1mm thick, 6.5mm diameter collimator is presented. Measurements on prototypes show good agreement with the designed characteristics.

In the context of dynamic contrast enhanced breast MR imaging we analyzed the effect of motion compensation by registration on the characterization of lesions. Two different registration techniques were applied: (1) rigid registration and (2) elastic registration based on the Navier-Lamé equation is applied. Interpreting voxels that exhibit a decline in image intensity after contrast injection (compared to the uncontrasted native image) as motion outliers, it can be shown that the rate of motion outliers can be largely reduced by both rigid and elastic registration. The performance of four lesion features including maximal signal enhancement ratio and variance of the signal enhancement ratio measured by area under the ROC curve as well as Cohen's κ value improved for elastic registration only, whereas features derived from rigidly registered images did not reach the performance level of either unregistered or elastically registered data.

C-arm based tomographic 3D imaging is applied in an increasing number of minimal invasive procedures. Due to the limited acquisition speed for a complete projection data set required for tomographic reconstruction, breathing motion is a potential source of artifacts. Intra-scan motion estimation and compensation is required. Here, a scheme for projection based local breathing motion estimation is combined with an anatomy adapted interpolation strategy and subsequent motion compensated filtered back projection. This approach is applied in animal experiments on a flat panel C-arm system delivering improved image quality in 3D liver tumor imaging.

The contrast of iodine to soft tissue (water) decreases with higher tube voltage in reconstructed 3D X-ray images. Improved acquisition protocols with a tube voltage of about 80 kV for imaging iodine have been proposed earlier for diagnostic CT imaging. We investigate the contrast-to-noise ratio (CNR) and the CNR-to-dose ratio (CDR) for different concentrations of iodinated contrast agent inserts in water background. The tube voltage of the protocol is lowered from 123 kV to 83 kV in 10 kV steps. A series of measurements with 16 different settings of tube voltage, current and filter settings are investigated. The weighted computed tomography dose index CTDIW for the new protocol settings is measured. Four protocols with tube voltages between 83 kV and 103 kV and similar X-ray dose are compared to the original protocol. A low contrast phantom, containing a water filled cylinder with 5 tubes of different mixtures of iodine contrast inside a 32 cm PMMA ring, is imaged with each protocol. Increased contrast of the iodine filled tubes to the water background is clearly visible in the reconstructed volumes for lower tube voltage and less copper filtering. The best results are obtained with the (83 kV, 561 mA, 0.4 Cu) – protocol. This protocol may improve iodine contrast agent visibility in various 3D imaging applications. For large patients a higher tube voltage, e.g. the (103 kV, 325 mA, 0.4 Cu) – protocol, may be used to avoid tube power limitations at 83 kV. This protocol still has improved iodine imaging compared to the 123 kV protocol and a larger tube power reserve.

Differential phase-contrast imaging in the x-ray domain provides three physically complementary pieces of information: the attenuation,the differential phase-contrast, related to the refractive index, and the dark-field signal, related to the total amount of radiation scattered into very small angles. In medical applications, it is of the utmost importance to present to the radiologist all clinically relevant information in as compact a way as possible. Hence, the needarisis for a method to combine two or more of the above mentioned images into one image containing all information relevant for diagnosis. We present an image composition algorithm that fuses the attenuation image and the differential phase contrast image into a composite image. The composition is performed in a noise optimal way such that the composite image is characterized by minimal noise-power at each frequency component.