Lothar Spies

Metal artifact reduction in CT using tissue-class modeling and adaptive prefiltering.

High-density objects such as metal prostheses, surgical clips, or dental fillings generate streak-like artifacts in computed tomography images. We present a novel method for metal artifact reduction by in-painting missing information into the corrupted sinogram. The information is provided by a tissue-class model extracted from the distorted image. To this end the image is first adaptively filtered to reduce the noise content and to smooth out streak artifacts. Consecutively, the image is segmented into different material classes using a clustering algorithm. The corrupted and missing information in the original sinogram is completed using the forward projected information from the tissue-class model. The performance of the correction method is assessed on phantom images. Clinical images featuring a broad spectrum of metal artifacts are studied. Phantom and clinical studies show that metal artifacts, such as streaks, are significantly reduced and shadows in the image are eliminated. Furthermore, the novel approach improves detectability of organ contours. This can be of great relevance, for instance, in radiation therapy planning, where images affected by metal artifacts may lead to suboptimal treatment plans.

Correction of scatter in megavoltage cone-beam CT

The role of scatter in a cone-beam computed tomography system using the therapeutic beam of a medical linear accelerator and a commercial electronic portal imaging device (EPID) is investigated. A scatter correction method is presented which is based on a superposition of Monte Carlo generated scatter kernels. The kernels are adapted to both the spectral response of the EPID and the dimensions of the phantom being scanned. The method is part of a calibration procedure which converts the measured transmission data acquired for each projection angle into water-equivalent thicknesses. Tomographic reconstruction of the projections then yields an estimate of the electron density distribution of the phantom. It is found that scatter produces cupping artefacts in the reconstructed tomograms. Furthermore, reconstructed electron densities deviate greatly (by about 30%) from their expected values. The scatter correction method removes the cupping artefacts and decreases the deviations from 30% down to about 8%.

Tomotherapeutic portal imaging for radiation treatment verification.

In their tomotherapy concept Mackie and co-workers proposed not only a new technique for IMRT but also an appropriate and satisfactory method of treatment verification. This method allows both monitoring of the portal dose distribution and imaging of the patient anatomy during treatment by means of online CT. This would enable the detection of inaccuracies in dose delivery and patient set-up errors. In this paper results are presented showing that a single electronic portal imaging device (EPID) could deliver all data necessary to establish such a complete verification system for tomotherapy and even other IMRT techniques. Consequently it has to be shown that it is able to record both the low-intensity photon fluences encountered in tomographic imaging and the intense photon transmission of each treatment field. The detector under investigation is a video-based EPID, the BIS 710 (manufactured by Wellhöfer Dosimetrie, Schwarzenbruck, Germany). To examine the suitability of the BIS for CT at 6 MV beam quality, different phantoms were scanned and reconstructed. The agreement between a diamond detector and BIS responses is quantitative. Tomographic reconstruction of a complete set of these transmission profiles resulted in images which resolve 3 cm large objects having a (theoretical) contrast to water of less than 9%. Three millimetre objects with a 100% contrast are clearly visible. The BIS signal was shown to measure photon fluence distributions. The reconstructed images possess a spatial and contrast resolution sufficient for accurate imaging of the patient anatomy, needed for treatment verification in many clinical cases.

Analytical scatter kernels for portal imaging at 6 MV.

X-ray photon scatter kernels for 6 MV electronic portal imaging are investigated using an analytical and a semi-analytical model. The models are tested on homogeneous phantoms for a range of uniform circular fields and scatterer-to-detector air gaps relevant for clinical use. It is found that a fully analytical model based on an exact treatment of photons undergoing a single Compton scatter event and an approximate treatment of second and higher order scatter events, assuming a multiple-scatter source at the center of the scatter volume, is accurate within 1% (i.e., the residual scatter signal is less than 1% of the primary signal) for field sizes up to 100 cm2 and air gaps over 30 cm, but shows significant discrepancies for larger field sizes. Monte Carlo results are presented showing that the effective multiple-scatter source is located toward the exit surface of the scatterer, rather than at its center. A second model is therefore investigated where second and higher-order scattering is instead modeled by fitting an analytical function describing a nonstationary isotropic point-scatter source to Monte Carlo generated data. This second model is shown to be accurate to within 1% for air gaps down to 20 cm, for field sizes up to 900 cm2 and phantom thicknesses up to 50 cm.

An iterative algorithm for reconstructing incident beam distributions from transmission measurements using electronic portal imaging.

The problem of reconstructing incident radiotherapy beam profiles from electronic portal images recorded behind a phantom is addressed. To this end an iterative algorithm is presented, which is able to extract the input beam profile from a portal image by compensating for the attenuation of the beam and subtracting the amount of scatter emitted by the phantom. The algorithm requires only a thickness map of the phantom. Scatter is estimated using a superposition method based on precalculated Monte Carlo scatter kernels. The method is tested for a homogeneous water-equivalent slab phantom for simple rectangular and complex multileaf collimated fields. It is shown that the method produces a stable result within four iterations yielding an accuracy for the incident beam distribution of better than 3%.

Segmentation-aided adaptive filtering for metal artifact reduction in radio-therapeutic CT images

In CT imaging, high absorbing objects such as metal bodies may cause significant artifacts, which may, for example, result in dose inaccuracies in the radiation therapy planning process. In this work, we aim at reducing the local and global image artifact, in order to improve the overall dose accuracy. The key part f this approach is the correction of the original projection data in those regions, which feature defects caused by rays traversing the high attenuating objects in the patient. The affected regions are substituted by model data derived from the original tomogram deploying a segmentation method. Phantom and climnical studies demonstrate that the proposed method significantly reduces the overall artifacts while preserving the information content of the image as much as possible. The image quality improvements were quantified by determining the signal-to-noise ratio, the artifact level and the modulation transfer function. The proposed method is computationally efficient and can easily be integrated into commercial CT scanners and radiation therapy planning software.

Fully Automated Atlas-Based Hippocampal Volumetry for Detection of Alzheimer's Disease in a Memory Clinic Setting

Hippocampal volume is a promising biomarker to enhance the accuracy of the diagnosis of dementia due to Alzheimers disease (AD). However, whereas hippocampal volume is well studied in patient samples from clinical trials, its value in clinical routine patient care is still rather unclear. The aim of the present study, therefore, was to evaluate fully automated atlas-based hippocampal volumetry for detection of AD in the setting of a secondary care expert memory clinic for outpatients. One-hundred consecutive patients with memory complaints were clinically evaluated and categorized into three diagnostic groups: AD, intermediate AD, and non-AD. A software tool based on open source software (Statistical Parametric Mapping SPM8) was employed for fully automated tissue segmentation and stereotactical normalization of high-resolution three-dimensional T1-weighted magnetic resonance images. Predefined standard masks were used for computation of grey matter volume of the left and right hippocampus which then was scaled to the patients total grey matter volume. The right hippocampal volume provided an area under the receiver operating characteristic curve of 84% for detection of AD patients in the whole sample. This indicates that fully automated MR-based hippocampal volumetry fulfills the requirements for a relevant core feasible biomarker for detection of AD in everyday patient care in a secondary care memory clinic for outpatients. The software used in the present study has been made freely available as an SPM8 toolbox. It is robust and fast so that it is easily integrated into routine workflow.

Fully Automated Atlas-Based Hippocampus Volumetry for Clinical Routine: Validation in Subjects with Mild Cognitive Impairment from the ADNI Cohort

Hippocampus volumetry based on magnetic resonance imaging (MRI) has not yet been translated into everyday clinical diagnostic patient care, at least in part due to limited availability of appropriate software tools. In the present study, we evaluate a fully-automated and computationally efficient processing pipeline for atlas based hippocampal volumetry using freely available Statistical Parametric Mapping (SPM) software in 198 amnestic mild cognitive impairment (MCI) subjects from the Alzheimers Disease Neuroimaging Initiative (ADNI1). Subjects were grouped into MCI stable and MCI to probable Alzheimers disease (AD) converters according to follow-up diagnoses at 12, 24, and 36 months. Hippocampal grey matter volume (HGMV) was obtained from baseline T1-weighted MRI and then corrected for total intracranial volume and age. Average processing time per subject was less than 4 minutes on a standard PC. The area under the receiver operator characteristic curve of the corrected HGMV for identification of MCI to probable AD converters within 12, 24, and 36 months was 0.78, 0.72, and 0.71, respectively. Thus, hippocampal volume computed with the fully-automated processing pipeline provides similar power for prediction of MCI to probable AD conversion as computationally more expensive methods. The whole processing pipeline has been made freely available as an SPM8 toolbox. It is easily set up and integrated into everyday clinical patient care.

Atlas based brain volumetry: How to distinguish regional volume changes due to biological or physiological effects from inherent noise of the methodology.

Fully-automated regional brain volumetry based on structural magnetic resonance imaging (MRI) plays an important role in quantitative neuroimaging. In clinical trials as well as in clinical routine multiple MRIs of individual patients at different time points need to be assessed longitudinally. Measures of inter- and intrascanner variability are crucial to understand the intrinsic variability of the method and to distinguish volume changes due to biological or physiological effects from inherent noise of the methodology. To measure regional brain volumes an atlas based volumetry (ABV) approach was deployed using a highly elastic registration framework and an anatomical atlas in a well-defined template space. We assessed inter- and intrascanner variability of the method in 51 cognitively normal subjects and 27 Alzheimer dementia (AD) patients from the Alzheimers Disease Neuroimaging Initiative by studying volumetric results of repeated scans for 17 compartments and brain regions. Median percentage volume differences of scan-rescans from the same scanner ranged from 0.24% (whole brain parenchyma in healthy subjects) to 1.73% (occipital lobe white matter in AD), with generally higher differences in AD patients as compared to normal subjects (e.g., 1.01% vs. 0.78% for the hippocampus). Minimum percentage volume differences detectable with an error probability of 5% were in the one-digit percentage range for almost all structures investigated, with most of them being below 5%. Intrascanner variability was independent of magnetic field strength. The median interscanner variability was up to ten times higher than the intrascanner variability.

Performance of prototype modules of a novel multislice CT detector based on CMOS photosensors

A novel CT detector based on CMOS photodiodes has been developed. A detector module comprises two identical photosensor arrays mounted to a ceramic substrate. Each sensor has a matrix of 20 by 10 pixels. Pixels are 1 mm (channel direction) x 1.8 mm (slice) large and consist of a photodiode, charge integration unit and a sample and hold stage. An automated switching between a low and a high sensitivity mode allows for a dynamic range of 17 bits. The integrated signals are read out, transferred to a printed circuit board (at a rate of 2463 Hz per pixel) and here converted into a digital data stream. The structured cadmium tungstate scintillator features lead stripes between pixels to reduce x-ray crosstalk and to shield the underlying in-pixel electronics. During assembling care was taken to ensure that the lead stripes of the scintillator entirely cover the pixel electronics underneath. Several prototype modules have been assembled and their performance concerning linearity, noise, crosstalk, and temperature dependence has been evaluated.