Computed Tomography: Image Principles

Image Format

• Images are reconstructed as an array of picture elements called pixels
• The matrix size refers to the number of pixels in the image
• Each pixel corresponds to a volume element, known as a voxel
• Slice thickness relates to the depth of a voxel

Principle of Computed Tomographic Imaging

• Narrow X-ray beams scan across a patient, in sync with a radiation detector on the opposite side
• For monoenergetic beams, X-ray transmission through the patient is expressed as: I = Ioe−μx
• Where I is the attenuated X-ray beam intensity, Io is the unattenuated beam intensity, and x is the material thickness
• For multiple regions (n) with varying linear attenuation coefficients, transmission is: I = Ioe−Σί=1μίχί
• Σί=1μixi = (μ1x1 + μ2x2 + ... + μnxn)

Image Reconstruction and Processing

• CT image reconstruction transforms X-ray penetration measurements into a digital image of a body section
• This process involves mathematical methods

Reconstruction Techniques

• Back projection
• Algebraic reconstruction
• Iterative reconstruction
• Filtered back projection

Backprojection

• Summed attenuation values are averaged across a row
• Multiple projections improve image quality
• Limitations include artefacts from too few projections
• Blurred images can be resolved via filtered backprojection
• Multislice scanners use filter interpolation, summing and averaging projections within axial slices

High Attenuation Example

• High attenuation circular objects on a matrix, shows each square represents a pixel
• The attenuation value is averaged per pixel
• Beams are projected in two directions, each beam obtains a summed linear attenuation coefficient (LAC)

Iterative Reconstruction

• Begin with Filtered backprojection to assign number values to all pixels in the matrix
• Computers then calculate expected detector readings from the generated image
• These calculations are compared to actual measurements, adjusting image values to match and refine the image
• Currently used almost exclusively, iterative reconstruction reduces CT dose, but calculations can be lengthy

CT Images

• CT images consist of pixels on a greyscale, where the grey level depends on the material density or linear attenuation coefficient
• Density is numerically represented using Hounsfield Units (HU), also known as CT numbers
• Hounsfield units are relative to water, which is defined as 0

Hounsfield Unit Values

• Bone: +1000 HU
• Liver: 40 to 60 HU
• White matter: 20 to 30 HU
• Grey matter: 37 to 45 HU
• Intravascular blood: 30 to 45 HU
• Fresh clotted blood: 70 to 80 HU
• Muscle: 10 to 40 HU
• Kidney: 30 HU
• CSF: 15 HU
• Water: 0 HU
• Fat: -50 to -100 HU
• Lung: -600 to -950 HU
• Air: -1000 HU

CT Number Window

• Hounsfield Units are visualized using an 8-bit greyscale with 128 levels
• Display settings include Window Level (WL), which is the CT number of mid-grey
• Window Width (WW) is the range of HU values displayed from black to white
• The ideal WW and WL depend on the clinical needs to visualize soft tissue, lung tissue, or bone
• The same image data can be displayed at different WL and WW settings