RC 1713 - Artefacts and pitfalls in tomography
1. To learn about the origins of image artefacts in tomographic imaging.
2. To understand image distortions in hybrid imaging.
3. To learn about solutions and workarounds.
A number of imaging techniques are currently present to facilitate clinical diagnosis. More specifically, CT, PET/CT and MR/PET are excellent imaging modalities for evaluation of a wide variety of pathologies, due to either their spatial resolution and/or tissue contrast. Furthermore, hybrid imaging can provide clinicians with tools for making an accurate diagnosis prior to making treatment recommendations. Their use, however, is occasionally restricted by a variety of associated artefacts. Artefacts degrade image quality leading in some cases in diagnostically useless images. In some cases, they also appear due to the inevitable consequence of natural properties of the human body. There are many different types of artefacts such as noise, beam hardening, scatter, pseudoenhancement, motion, cone beam, helical, ring or metal artefacts. Understanding the technical basics of hybrid imaging is vital to avoid misinterpreting artefacts as pathological findings. In addition, a deeper understanding of the origins and imaging characteristics of artefacts will reduce the likelihood of misdiagnosis and optimise the image. As accurate CT, PET/CT and MR/PET imaging is a question of knowledge, the particular refresher course intends to refresh the participants’ knowledge of the physical origins of artefacts. The course will present different methods for minimising artefacts, how to image regions of the body close to metal implants, as well as solutions to frequent image distortion.
1. To understand the source of artefacts in clinical CT.
2. To understand the most important correction methods.
3. To find out what artefact correction techniques are actually provided by the CT vendors in their systems.
Although CT is the most quantitative diagnostic tomographic imaging modality, its images still suffer from several kinds of artefacts. Among the CT artefacts the most prominent, severe, and well known one in diagnostic CT is the metal artefact, mainly because larger metal implants are almost opaque to the x-ray radiation and thus x-rays that are scattered from the surrounding tissue into the metal shadow cause a very high scatter-to-primary ratio. In addition, there are many less dominating sources of artefacts. Among those are beam hardening and scatter causing dark streaks between dense objects such as bones, motion causing motion blurring and partial cycloid artefacts, very large patients causing truncation artefacts, sampling issues causing aliasing artefacts, as well as the finite detector size which causes linear and non-linear partial volume artefacts. Last but not least, there are artefacts that are known mainly to experts in CT physics because the manufacturers typically correct for them: defect detector pixel artefacts, detector afterglow artefacts, and geometric misalignment artefacts. The lecture discusses the source of artefacts and gives, wherever applicable, examples and points towards approaches of how to reduce such artefacts.
1. To understand image distortions, artefacts and bias from methodological pitfalls.
2. To appreciate and understand solutions to frequent image distortions.
3. To understand the methodological limitations of PET/CT.
A proposal to combine PET with CT was made in the early 1990s by Townsend, Nutt et al. In addition to the intrinsic alignment of complementary images, the anticipated benefit of PET/CT was to use the CT images to derive the mandatory attenuation maps for the PET data. In short, CT-based attenuation correction (CT-AC) is based on the assumption that CT image can be segmented into bone and non-bone tissues; voxels in each tissue class are then scaled with corresponding scale factors. CT-AC is prone to several errors arising from the methodological shortcomings of the segmentation-scaling method in light of CT-transmission measurements in clinical conditions. These include: truncation artefacts, artefacts from high-density implants and positive contrast agents and others. More frequently, errors from patient motion during the examination propagate through CT-AC into the final emission images and lead to distortions/bias of the reconstructed data. In addition, artefacts and biases may occur from involuntary mistakes made during the set-up/conduct of the imaging procedure. During this presentation, we will rehearse the principles of CT-AC in PET/CT and point to source of artefacts arising from the methodology of CT-AC and from specific imaging workflow scenarios not optimized for routine PET/CT.
1. To identify common artefacts.
2. To understand the physical origin of and methods to resolve artefacts.
3. To understand the interrelation of MR artefacts and bias in PET quantification.
Each new imaging modality and technical system introduces new types of artefacts and in MR/PET hybrid imaging the potential for new artefacts is even higher than just considering two independent systems. Artefacts in MR/PET might affect the visual impression of either MR or PET data and, furthermore, may also have an effect on quantification in MR and even stronger on PET being a quantitative method. Artefacts in integrated MR/PET may result from technical crosstalk between the MR and the PET components. Both imaging centres might not be co-aligned correctly. Differences in the data acquisition speed between MR and PET might lead to local misalignments due to patient motion. MR-based attenuation correction (AC) is still a new concept to support PET data quantification in the best possible way. All deviations from the real physical gamma quanta attenuation will ultimately lead to false values in PET quantification following AC. Administration of contrast agents before application of MR-based AC due to changes in tissue contrast potentially may lead to errors in tissue segmentation. The MR field-of-view is limited which may lead to truncation of patient tissues exceeding the constraints of the field-of-view. Consequently, the arms are not fully considered in AC leading to false PET quantification. Metal implants might introduce signal voids or local distortions in MR-AC that exceed the physical implant volume. Such signal voids might then be assigned with the low linear attenuation coefficients of air in image segmentation. Typical artefacts, pitfalls, and their avoidance will be presented.