Metal artifact reduction for CT-based luggage screening.

BACKGROUND: In aviation security, checked luggage is screened by computed tomography scanning. Metal objects in the bags create artifacts that degrade image quality. Though there exist metal artifact reduction (MAR) methods mainly in medical imaging literature, they require knowledge of the materials in the scan, or are outlier rejection methods. OBJECTIVE: To improve and evaluate a MAR method we previously introduced, that does not require knowledge of the materials in the scan, and gives good results on data with large quantities and different kinds of metal. METHODS: We describe in detail an optimization which de-emphasizes metal projections and has a constraint for beam hardening and scatter. This method isolates and reduces artifacts in an intermediate image, which is then fed to a previously published sinogram replacement method. We evaluate the algorithm for luggage data containing multiple and large metal objects. We define measures of artifact reduction, and compare this method against others in MAR literature. RESULTS: Metal artifacts were reduced in our test images, even for multiple and large metal objects, without much loss of structure or resolution. CONCLUSION: Our MAR method outperforms the methods with which we compared it. Our approach does not make assumptions about image content, nor does it discard metal projections.

Flexible methods for segmentation evaluation: results from CT-based luggage screening.

BACKGROUND: Imaging systems used in aviation security include segmentation algorithms in an automatic threat recognition pipeline. The segmentation algorithms evolve in response to emerging threats and changing performance requirements. Analysis of segmentation algorithms' behavior, including the nature of errors and feature recovery, facilitates their development. However, evaluation methods from the literature provide limited characterization of the segmentation algorithms. OBJECTIVE: To develop segmentation evaluation methods that measure systematic errors such as oversegmentation and undersegmentation, outliers, and overall errors. The methods must measure feature recovery and allow us to prioritize segments. METHODS: We developed two complementary evaluation methods using statistical techniques and information theory. We also created a semi-automatic method to define ground truth from 3D images. We applied our methods to evaluate five segmentation algorithms developed for CT luggage screening. We validated our methods with synthetic problems and an observer evaluation. RESULTS: Both methods selected the same best segmentation algorithm. Human evaluation confirmed the findings. The measurement of systematic errors and prioritization helped in understanding the behavior of each segmentation algorithm. CONCLUSIONS: Our evaluation methods allow us to measure and explain the accuracy of segmentation algorithms.

Segmentation of artifacts and anatomy in CT metal artifact reduction.

PURPOSE: Metal objects present in x-ray computed tomography (CT) scans are accompanied by physical phenomena that render CT projections inconsistent with the linear assumption made for analytical reconstruction. The inconsistencies create artifacts in reconstructed images. Metal artifact reduction algorithms replace the inconsistent projection data passing through metals with estimates of the true underlying projection data, but when the data estimates are inaccurate, secondary artifacts are generated. The secondary artifacts may be as unacceptable as the original metal artifacts; therefore, better projection data estimation is critical. This research uses computer vision techniques to create better estimates of the underlying projection data using observations about the appearance and nature of metal artifacts. METHODS: The authors developed a method of estimating underlying projection data through the use of an intermediate image, called the prior image. This method generates the prior image by segmenting regions of the originally reconstructed image, and discriminating between regions that are likely to be metal artifacts and those that are likely to represent anatomical structures. Regions identified as metal artifact are replaced with a constant soft-tissue value, while structures such as bone or air pockets are preserved. This prior image is reprojected (forward projected), and the reprojections guide the estimation of the underlying projection data using previously published interpolation techniques. The algorithm is tested on head CT test cases containing metal implants and compared against existing methods. RESULTS: Using the new method of prior image generation on test images, metal artifacts were eliminated or reduced and fewer secondary artifacts were present than with previous methods. The results apply even in the case of multiple metal objects, which is a challenging problem. The authors did not observe secondary artifacts that were comparable to or worse than the original metal artifacts, as sometimes occurred with the other methods. The accuracy of the prior was found to be more critical than the particular interpolation method. CONCLUSIONS: Metals produce predictable artifacts in CT images of the head. Using the new method, metal artifacts can be discriminated from anatomy, and the discrimination can be used to reduce metal artifacts.

Validation of radiographic simulation codes including x-ray phase effects for millimeter-size objects with micrometer structures.

The mix between x-ray phase and attenuation information needs to be understood for accurate object recovery from radiography and tomography data. We are researching and experimentally validating algorithms that simulate x-ray phase contrast to determine the required physics necessary for quantitative object recovery. The results of a study are described to determine if a multislice (beam-propagation) method is required for simulating x-ray radiographs. We conclude that the multislice method is not required for accurate simulation of greater than or equal to 8 keV x-ray radiographs of millimeter-size objects with micrometer structures.

Using exact Poisson likelihood functions in Bayesian interpretation of counting measurements.

A technique for computing the exact marginalized (integrated) Poisson likelihood function for counting measurement processes involving a background subtraction is described. An empirical Bayesian method for determining the prior probability distribution of background count rates from population data is recommended and would seem to have important practical advantages. The exact marginalized Poisson likelihood function may be used instead of the commonly used Gaussian approximation. Differences occur in some cases of small numbers of measured counts, which are discussed. Optional use of exact likelihood functions in our Bayesian internal dosimetry codes has been implemented using an interpolation-table approach, which means that there is no computation time penalty except for the initial setup of the interpolation tables.