CT Image Quality and Parameters
This elearning is designed for the active technologist performing CT scans in the clinical routine. Discussed concepts apply to everyday’s work and help understand the meaning of various components of a scan protocol and how changes of CT parameters influence image quality
Welcome to this web based training on CT Image Quality and Parameters. In this course, we will be discussing three main topics: Criteria for CT Image Quality Influences on CT Image Quality Definition of Image Display Explain the different criteria of image quality in terms of what they are and how they are defined. By the end of this course, you will be able to: Explain CT parameters and their impact on image quality List which CT parameters are used to adjust for better image quality pre and post-acquisition. Identify dose optimization tools and how they impact image quality. High/Spatial Resolution Low Contrast Resolution Temporal Resolution Image Definition Noise Artifacts The ability to define small objects and details and be able to differentiate them from surrounding structures with a very high density difference. The ability to differentiate objects and details from surrounding structures that have very little density difference. Ability to visualize adjacent tissues that are similar in densities Image noise less tolerated; higher dose required Also called "time resolution", temporal resolution is the amount of time it takes to acquire the image (exposure time). The shorter the acquisition, the higher the temporal resolution and the less the motion artifacts. Good temporal resolution is essential for cardiac studies. Image definition means the sharpness of an object relative to surrounding tissue. It depends on: Rotation time Image display Operating mode Kernel Slice Width Image "noise" is determined by the number of x-ray quanta that reach the detector and then contribute to the image. The greater the quanta (higher dose), the lower the noise. Noise also depends on: mAs kV Kernel Slice width Patient size Collimation In the example shown here, you can see that the image on the right is more grainy, or noisy. Noise here was caused by the use of low mA. Artifacts are the various structures or patterns that appear in a CT image, but are not found in the original object. Examples of parameters that may cause artifacts: Scan time (motion) Operating mode (Beam Hardening) Slice width (Partial Volume) System (Rings) Patient (Metal) Image artifacts may appear as: Streaks Dark Bars Rings Motion artifacts can be compensated for by the Motion Correction Algorithm (MCA) w/o correction w/ correction . Metals such as gold absorb x-radiation almost completely which produces "radiation shadows". These lead to pronounced streak artifacts over the entire reconstructed image. These artifacts can be avoided via a gantry tilt to exclude the metallic object from the direct slice plane. Normally, CT values can be measured from -1024 to +3071 Hounsfield Units. This can be extended to -10240 to +30710 to visualize metals of high attenuation. Note: This does NOT reduce streak artifacts. This image shows an evaluation of a post-operative femoral hip replacement for metal fracture. CT value is 6000 HU. Streak-like artifacts, also called partial volume artifacts, occur most frequently in the bony structures at the base of the skull and the petrous bone region. This is because the very dense structures (bones) are only partially included in the slice, resulting in high contrast errors. Selecting a thinner slice prevents such artifacts from occuring since high contrast structures are partially inluded less frequently. However, this inherently increases the noise level, thus degrading low contrast resolution. Combines several thin slices, reducing the partial volume artifact Provides a thicker slice, reducing pixel noise and offering good soft tissue discrimination The x-ray photons emitted from the x-ray tube do not all have the same energy. As they penetrate the object, the attenuation of low energy photons is in the middle of the object, resulting in an increase in the x-ray energy at the center of the object. The spectrum is shifted to higher energies called beam hardening. In the image, streak artifacts or the "cupping effect" can be seen. w/o correction w/ correction More Streaking Less Streaking w/o correction w/ correction Severe Cupping Homogeneous CT Values X-ray attenuation is greater in the lateral position than along the anterior position, therefore directional noise is seen. w/o A.F. w/ A.F. The correct organ program mode must be selected. There are hidden values that correspond to individual body parts (e.g., adaptive filter, motion correction) Identify poor image quality and technical defects caused by the CT scanner Adjust and manipulate scan protocol parameters Dose optimization methods CT Image Quality System User Patient The individual detector elements of a detector system may not produce the same signal for the same irradiation. When a detector element puts out an erroneous signal, ring artifacts appear. This can be eliminated by calibration, if not call technical service. Pre-Calibration Post-Calibration Streak Artifacts CT Image Quality mAs Rotation Time Collimation Slice Thickness/Width Patient kV Kernel · Absorbed dose: the radiation energy transferred to an anatomic structure during irradiation of that structure (as long as kV is not changed) · Soft tissue diagnosis (liver, brain) · Low contrast resolution - requires small structures to be well discriminated · CT values of surrounding tissue are very close. · Noise must be kept to a minimum. · Higher mAs is necessary. · High Contract structures (bone, lung) · High contrast resolution - produce CT values which are very high compared to surrounding tissues. · Requires lower mAs. Noise ~ 1/sqrt (mAs) 25 mAs 100 mAs Image 1: Image 2: Low mAs value, high noise 4x mAs value, half noise Effective mAs = mAs/Pitch Factor Noise ~ 1/sqrt (mAs) Image Noise Effective mAs Killivoltage (kV) determines the energy level or penetrating power of the x-ray beam. It affects the patient dose, image noise, and contrast. Because the attenuation of tissues is dependent on the kV setting to varying degrees, the tube voltage also influences the CT numbers, Hounsfield Units (HU), in the image. Tube voltage has a greater influence on patient's dose than mAs. Increasing the kV causes the dose to increase as shown in this graph. Relative Dose kV Imaging at higher kV Reduced image noise Decreased image contrast Increased dose Imaging at lower kV Reduced dose Increased signal-to-noise ratio (CTA) The shortest possible rotation time that produces sufficient dose should be employed to reduce motion artifacts. 0.75 seconds 0.33 seconds The convolution Kernel is a reconstruction parameter affecting image sharpness and noise. The kernel applies a specific mathematical algorithm that digitally filters the raw data during image reconstruction. Sharp Image Definition Smooth Noise Types of Kernels H - Head B - Body U - Ultra High Resolution C - Child Head S - Special Applications D - Dual Energy Examples: B30f - Body Kernel 30 fast mode H70s - Head kernel 70 standard mode Image Appearance Typical Kernal Numbers Visualization/Purpose Smooth 10-20 3D post-processing, noise reduction with thin slices Medium 30-50 General soft tissue display Sharp 60-70 Lung or bone visualization with edge enhancement High Resolution 80-90 High spatial resolution assessment of minute structure, e.g. inner ear Sharp kernels provide better spatial resolution, but with more noise. Medium Kernel Sharp Kernel Routine brain study with smooth kernel H31s and soft window Routine abdomen study with smooth kernel B30 and soft window Ankle study with sharp kernel U80 and bone window Lung examination with sharp kernel B70 and lung window Determines the slice or section thickness that will be utilized for a particular CT scanning procedure. Determines the minimum slice width that can be used for image reconstruction. Narrower reconstructed slice width results in improved axial resolution, but also increases noise. User determines the tube collimation by narrowing or widening the x-ray beam. Also called slice collimation NOT slice width or slice thickness, which is the thickness of the reconstructed image. Slice width and slice collimation used to be regarded as synonyms for single-slice CT imaging because only images with a width equal to scan collimation could be reconstructed. With Multi-Slice CT slice width does not equal slice collimation so the terms are not synonymous. Narrow collimation, 64 slices x 0.6 mm Wide collimation, 16 slices x 1.2 mm Selecting a suitable slice width or thickness requires a balance between edge definition and noise because of their mutually offsetting effects. A thick slice means: low noise better low constrast resolution less images, easier storage poorer edge definition partial volume artifacts recommeded for regular soft tissue studies with 2D post-processing A thin slice means: high noise poorer low contrast resolution more images better edge definition better high contrast resolution less partial volume artifact recommended for detailed anatomical structures and 3D presentations 3 mm Slice 10 mm Slice 0.75 mm Slice 5 mm Slice Thinner slices give better spatial resolution for bony structures, hence better image detail. 5 mm Slice 1 mm Slice Thinner Slice Thickness - 0.75 mm Thicker Slice Thickness - 5.0 mm Patient is positioned in the isocenter – optimal dose and image quality Patient is positioned too high – increased mAs Patient is positioned too low – reduced mAs and increased noise Noise Patient Diameter As Low Reasonably As Achievable Rule of Thumb: The noise level doubles for every 8 cm increase in patient diameter. Automatic Exposure Control (AEC) Fully automated dose management system Every patient is unique Anatomy based No user interaction Scan with constant mA Reduced dose level based on topogram Real time angular dose modulation Dose adaptation settings Effective mAs (implemented in scan protocol) Average: default Strong Less dose to slim patients, more noise More dose to obese patients, less noise Weak More dose to slim patients, less noise Less dose to obese patients, more noise Quantity Description Meaning Unit CTDIvol Average dose over the total volume scanned for the selected CT conditions of operation Basic dose parameter in CT mGy DLP Product of the CTDIvol and the scan range Main descriptor of the total energy deposited in the body mGy x cm Effective Dose Average dose to the whole body, which is the weighted average of all affected organs Describes the radiation risk mSv Field of View (FoV) Resizing FoV using raw data vs. magnification tool Windowing Houndsfield Units (HU) Width/Center Image filters vs. kernels Reconstruction increment Lung Window Mediastinum Window Rule of Thumb: The CT value of water is 0 and air is -1000. The relative values of the other tissues are calculated relative to that of water. Bone Water Fat Air Center Modify image brightness Set to average Hounsfield Unit for investigated tissue Soft Tissue: 30 to 50 HU Lung: -400 to -600 HU Bone: 300 to 500 HU Width Cover all densities while allowing tissue discrimination Modify image contrast Hounsfield Unit CT Windowing W80/C40 W500/C-1500 W2000/500 Narrow window width for white and gray matter differentiation Wider window width and higher center for bone density Narrow window width for better contrast visualization/lesion detection Wider window width to look at soft tissue (i.e. fat) Reconstruction: A reconstruction from raw data enhances sharpness of details. Magnification: A purely optical magnification of image data which may result in blurred appearance. Post-acquisition procedure to correct inadequate image quality: Use of inappropriate kernel Create new image impression Raw data no longer available Filter Name Filter Type Results/Features LCE - Low Contrast Enhancement Smoothing 4 grades: small, medium, strong, very strong Less noise, better low-contrast resolution and tissue differentiation. Used mainly for soft tissue imaging. Similar results to smooth kernel utilization. HCE - High Contrast Enhancement Sharpening Better high-contrast resolution. Sharper but grainier image. Used for detailed structures: mainly bones and lungs with appropriate window settings. Similar results to sharp kernel utilization. ASA - Advanced Smoothing Algorithm Smoothing A numeric value is required. The higher the value the lower the noise. Default values available for different organs. Similar to LCE filter. Reduces image noise without loss of sharpness. Low Contrast Enhancement Filter-LCE High Contrast Enhancement Filter-HCE Left: Without Automatic Bone Correction Filter Right: With Automatic Bone Correction Filter Filters Manipulation of image pixels, not raw data Possible to achieve similar image impression Limited resolution Solution when raw data no longer available Kernels Manipulation of raw data Full resolution Method of choice Three consecutive 5 mm slices with 50% overlap Two contiguous 5 mm slices reconstructed without overlap You should now be able to: Explain the different criteria of image quality in terms of what they are and how they are defined. Explain CT parameters and their impact on image quality. List which CT parameters are used to adjust for better image quality pre and post acquisition. Identify dose optimization tools and how they impact image quality 360° rotation of the tube-detector Determines speed and temporal resolution Shorter rotation times Routine chest, vascular and pediatric studies Longer rotation times. Routine brain and lumbar spine examinations As seen in the graph here, the shorter the rotation time, the less likely motion artifacts occur. The longer the rotation time, the more likely motion artifacts may arise. Motion Scan Time (s) Tube current (mA) and scan time determine dose. Noise mAs Rule of Thumb: The higher the dose, the lower the noise.