Clemens Maaß

Image-based dual energy CT using optimized precorrection functions: A practical new approach of material decomposition in image domain

Dual energy CT (DECT) measures the object of interest using two different x-ray spectra in order to provide energy-selective CT images or in order to get the material decomposition of the object. Today, two decomposition techniques are known. Image-based DECT uses linear combinations of reconstructed images to get an image that contains material-selective DECT information. Rawdata-based DECT correctly treats the available information by passing the rawdata through a decomposition function that uses information from both rawdata sets to create DECT specific (e.g., material-selective) rawdata. Then the image reconstruction yields material-selective images. Rawdata-based image decomposition generally obtains better image quality; however, it needs matched rawdata sets. This means that physically the same lines need to be measured for each spectrum. In todays CT scanners, this is not the case. The authors propose a new image-based method to combine mismatched rawdata sets for DECT information. The method allows for implementation in a scanners rawdata precorrection pipeline or may be used in image domain. They compare the ability of the three methods (image-based standard method, proposed method, and rawdata-based standard method) to perform material decomposition and to provide monochromatic images. Thereby they use typical clinical and preclinical scanner arrangements including circular cone-beam CT and spiral CT. The proposed method is found to perform better than the image-based standard method and is inferior to the rawdata-based method. However, the proposed method can be used with the frequent case of mismatched data sets that exclude rawdata-based methods.

Exact dual energy material decomposition from inconsistent rays (MDIR).

PURPOSE Dual energy CT (DECT) allows calculating images that show the spatial distribution of the electron density and the atomic number or, more common, images of two basis material densities. In contrast, the Hounsfield unit that is shown in standard CT images is a measure of the x-ray attenuation, which is a function of the atomic number and electron density. To acquire additional information, DECT measures the object of interest using two different detected x-ray spectra. Most clinical CT scanners realize dual energy CT by fast tube voltage switching or by dual source dual detector arrangements and therefore do not allow measuring geometrically identical lines with each spectrum. Then, it is not possible to preprocess the raw data and calculate dual energy-specific raw data sets. The combination of the information of both spectra rather needs to be carried out in image domain after image reconstruction. Compared to the ideal raw data-based dual energy approaches, those image-based DECT methods are inferior because they are not able to correctly deal with the polychromatic nature of the x-rays. This article proposes a dedicated dual energy reconstruction algorithm for inconsistent rays that correctly accounts for all spectral effects. METHODS Material decomposition from inconsistent rays (MDIR) is an iterative method to indirectly perform raw data-based DECT even though different lines were measured for both spectra. Its iterative nature allows treating the x-ray polychromaticity correctly. The iterative process is initialized by density images that were obtained from an image-based material decomposition. Those images suffer from errors that originate from the polychromatic nature of the spectra. These errors are calculated by polychromatic forward projection of each measured line. After correction of the initial material density images, the polychromatic forward projection is repeated with more accurate material density images, yielding a more accurate error calculation. To demonstrate the proposed method, simulations and measurements were performed using clinical and preclinical dual source dual energy CT scanners. RESULTS Two iterations of MDIR are sufficient to greatly improve the qualitative and quantitative information in material density images. It is shown that banding artifacts, cupping artifacts, and mean density errors can be completely eliminated. Simulations with high geometrical inconsistency between the rays of different spectra indicate that nearly exact material decomposition is possible with MDIR. Furthermore, simulations show that the method works well in the presence of materials with K-edges within the detected spectrum. Phantom measurements using a clinical dual source CT scanner show the elimination of artifacts, which cause up to 4% mean density error. CONCLUSIONS At moderate computational burden, the proposed MDIR algorithm yields images of the same high quality as direct raw data-based DECT methods. In contrast to those, MDIR is applicable to the case of inconsistent rays, as it often occurs in clinical or preclinical CT. Compared to image-based methods MDIR reduces artifacts and improves mean density errors in material density images. All dual energy postprocessing methods that are in use today, such as bone removal, virtual noncontrast images, etc., can be applied to the images provided by MDIR.

A new weighting function to achieve high temporal resolution in circular cone-beam CT with shifted detectors.

The size of the field of measurement (FOM) in computed tomography is limited by the size of the x-ray detector. In general, the detector is mounted symmetrically with respect to the rotation axis such that the transaxial FOM diameter approximately equals the lateral dimensions of the detector when being demagnified to the isocenter. To enlarge the FOM one may laterally shift the detector by up to 50% of its size. Well-known weighting functions must then be applied to the raw data prior to convolution and backprojection. In this case, a full scan or a scan with more than 360° angular coverage is required to obtain complete data. However, there is a small region, the inner FOM, that is covered redundantly and where a partial scan reconstruction may be sufficient. A new weighting function is proposed that allows one to reconstruct partial scans in that inner FOM while it reconstructs full scan or overscan data for the outer FOM, which is the part that contains no redundancies. The presented shifted detector partial scan algorithm achieves a high temporal resolution in the inner FOM while maintaining truncation-free images for the outer part. The partial scan window can be arbitrarily shifted in the angular direction, what corresponds to shifting the temporal window of the data shown in the inner FOM. This feature allows for the reconstruction of dynamic CT data with high temporal resolution. The approach presented here is evaluated using simulated and measured data for a dual source micro-CT scanner with rotating gantry.

New approaches to region of interest computed tomography.

PURPOSE In classical x-ray CT, the diameter of the field of measurement (FOM) must not fall below the transversal diameter of the patient or specimen. Thereby, the ratio of the diameter of FOM and the number of transversal detector elements typically defines the spatial resolution. The authors aim at improving the spatial resolution within a region of interest (ROI) by a factor of 10-100 while maintaining artifact-free CT image reconstruction inside and outside the ROI. Two novel methods are proposed for artifact-free reconstruction of the truncated ROI scan (data weighting method and data filtering method) and compared with the gold standard (data completion method) for this problem. METHODS First, an overview scan with low spatial resolution and a large FOM that exceeds the object transversally is performed. Second, a high-resolution scan is performed, where the scanners magnification is changed such that the FOM matches the ROI at the cost of laterally truncated projection data. The gold standard is forward projecting the low-resolution scan on the rays missing in the high-resolution scan. The authors propose the data filtering method, which uses the low-resolution reconstruction and calculates a high frequency correction term from the high-resolution scan, and the data weighting method, which reconstructs the truncated high-resolution data and calculates a detruncation image from the low-resolution data. RESULTS The methods are compared using a simulation of the Forbild head phantom and a measurement of a spinal disk implant. The results of the data weighting method and the data completion method show the same image quality. The data filtering method yields slightly inferior image quality that may still be sufficient for many applications. Both new methods considerably outperform the data completion method regarding the computational load. CONCLUSIONS The new ROI reconstruction methods are superior to the gold standard regarding the computational load. Comparing the image quality with the gold standard, the data filtering method is slightly inferior and the data weighting method yields equal quality.

Empirical scatter correction (esc): A new CT scatter correction method and its application to metal artifact reduction

Scatter artifacts impair the CT image quality and the accuracy of CT values. Especially in cases with metal implants and in wide cone-angle flat detector CT scans, scatter artifact removal can be of great value. Typical scatter correction methods try to estimate scattered radiation and subtract the estimated scatter from the uncorrected data. Scatter is found either by time-consuming Monte Carlo-based simulations of the photon trajectories, or by rawdata-based modelling of the scatter content using scatter kernels, whose open parameters have to be determined very accurately and for each scanner and type of object individually, and that sometimes even require a data base of typical objects. The procedures are time-consuming and require intimate knowledge about the scanner, in particular about the spectral properties, for which a correction is designed. We propose an empirical scatter correction (ESC) algorithm which does not need lots of prior knowledge for calibration. ESC assumes that a linear combination of the uncorrected image with various ESC basis images is scatter-free. The coefficients for the linear combination are determined in image domain by maximizing a flatness criterion of the combined volume. Here, we minimized the total variation in soft tissue regions using the gradient descent method with a line search. Simulated data and several patient data sets acquired with a clinical cone-beam spiral CT scanner, where scatter was added using a Monte Carlo scatter calculation algorithm, were used to evaluate ESC. Metal implants were simulated into those data sets, too. Our preliminary results indicate that ESC has the potential to efficiently reduce scatter artifacts in general, and metal artifacts in particular. ESC is computationally inexpensive, highly flexible, and does not require know-how of the scanner properties.

Simple ROI cone-beam computed tomography

In classical x-ray CT the diameter of the field of measurement (FOM) must not fall below the transversal diameter of the patient or specimen. Thereby, the ratio of the the diameter of the FOM and the number of transversal detector elements typically defines the spatial resolution. We aim to improve the spatial resolution within a region of interest (ROI) by a factor of 10 to 100 while maintaining artifact-free CT image reconstruction inside and outside the ROI. Two novel methods are proposed for artifact-free reconstruction of the truncated ROI scan (data weighting method and data filtering method) and compared with the gold standard (data completion method) for this problem. First, an overview scan with low spatial resolution and a large FOM that exceeds the object transversally is performed. Second, a high resolution scan is performed where the scanners magnification is changed such that the FOM matches the ROI at the cost of laterally truncated projection data. The gold standard is forward projecting the low resolution scan on the rays missing in the high resolution scan. We propose the data filtering method, which uses the low resolution reconstruction and calculates a high frequency correction term from the high resolution scan, and the data weighting method, which reconstructs the truncated high resolution data and calculates a detruncation image from the low resolution data. The methods are compared using a simulation of the Forbild head phantom and a measurement of a spinal disk implant. The results of the data weighting method and the data completion method show the same image quality. The data filtering method yields inferior image quality because artifacts (partial volume effect, noise) of the overview scan propagate into the ROI reconstruction. Both new methods considerably outperform the data completion method regarding the computational load. The new ROI reconstruction methods are superior to the gold standard regarding the computational load. Comparing the image quality with the gold standard, the data filtering method is inferior and the data weighting method yields equal quality.

Quantification of temporal resolution and its reliability in the context of TRI-PICCS and dual source CT

Temporal resolution is an important issue especially in cardiac CT. To quantify it, often merely the time that is needed to acquire rawdata that contribute to a reconstructed image is used. In combination with more complex reconstruction algorithms, which aim to improve the temporal resolution, (e.g. TRI-PICCS) this procedure has proven to be inadequate. This study proposes and evaluates a more accurate simulation-based technique to assess the temporal resolution of a CT system (including its reconstruction algorithm). To calculate the temporal resolution of the system on a single point within the field of measurement, a vessel which performs a cardiac motion pattern is simulated at this position. The motion pattern is adapted such that the accuracy loss caused by motion exactly meets a defined threshold and then the temporal resolution can be taken from that motion pattern. Additionally the dependency of the temporal resolution on the direction of the motion is evaluated to obtain a measure of the reliability. The method is applied to single source and dual source full scan and short scan reconstructions as well as on TRI-PICCS reconstructions. The results give an accurate impression on the system response to motion. In conclusion, the proposed method allows quantifying the temporal resolution of a CT system as a function of many parameters (motion direction, source position, object position, reconstruction algorithm).

Comparing short scan CT reconstruction algorithms regarding cone-beam artifact performance

While circular cone-beam computed tomography (CBCT) has numerous advantages in CT scanner design, the image reconstruction from circular cone-beam data is known to be approximate since the data sufficiency condition is not fulfilled. This is an unresolved issue yet and may become a problem if the number of detector rows is further increased in the future. The purpose of this work is to evaluate different CBCT reconstruction algorithms regarding their cone-beam artifact performance. For the comparison the two standard methods Feldkamp and SART are used. Additionally the factorization approach, which is designed for reconstruction with low cone-beam artifacts and a new method for cone-beam artifact reduction (CBAR) based on iterative filtered backprojection reconstruction are used. Among the compared algorithms, the Feldkamp reconstruction has the lowest computational load and shows the worst cone-beam artifacts while the factorization approach shows the best trade-off between cone-beam artifacts and computation time.

TRI-PICCS in single source and dual source CT

Recently, temporal resolution improved prior image constrained compressed sensing (TRI-PICCS), an algorithm promising improved temporal resolution in cardiac CT, was proposed. Here, we extend the ideas to dual source CT. Single source and dual source CT simulations of a heart phantom as well as a single source cardiac CT measurement are used to evaluate the temporal resolution of CT images reconstructed with TRI-PICCS. Short scan reconstructions of the same data are used for comparison. The result of this study is that there seems to be an advantage in temporal resolution when using TRI-PICCS, however, this impression is actually caused by a slight shift of the motion phase when using TRI-PICCS. Comparing short scan reconstructions and TRI-PICCS reconstructions at the same motion phase, no temporal resolution improvement could be confirmed.

CT image reconstruction with half precision floating‐point values

PURPOSE Analytic CT image reconstruction is a computationally demanding task. Currently, the even more demanding iterative reconstruction algorithms find their way into clinical routine because their image quality is superior to analytic image reconstruction. The authors thoroughly analyze a so far unconsidered but valuable tool of tomorrows reconstruction hardware (CPU and GPU) that allows implementing the forward projection and backprojection steps, which are the computationally most demanding parts of any reconstruction algorithm, much more efficiently. METHODS Instead of the standard 32 bit floating-point values (float), a recently standardized floating-point value with 16 bit (half) is adopted for data representation in image domain and in rawdata domain. The reduction in the total data amount reduces the traffic on the memory bus, which is the bottleneck of todays high-performance algorithms, by 50%. In CT simulations and CT measurements, float reconstructions (gold standard) and half reconstructions are visually compared via difference images and by quantitative image quality evaluation. This is done for analytical reconstruction (filtered backprojection) and iterative reconstruction (ordered subset SART). RESULTS The magnitude of quantization noise, which is caused by a reduction in the data precision of both rawdata and image data during image reconstruction, is negligible. This is clearly shown for filtered backprojection and iterative ordered subset SART reconstruction. In filtered backprojection, the implementation of the backprojection should be optimized for low data precision if the image data are represented in half format. In ordered subset SART image reconstruction, no adaptations are necessary and the convergence speed remains unchanged. CONCLUSIONS Half precision floating-point values allow to speed up CT image reconstruction without compromising image quality.