CT Imaging Techniques and Principles

Image Reconstruction and Manipulation

• Algorithms are sets of rules for getting a specific output from an input.
• Fourier transform describes the amplitude and phase of each sinusoid, converting spatial domain signals to frequency domain signals.
• Convolution is a digital image processing technique using a filter function to modify images.
• Interpolation is a mathematical technique to estimate a function's value from known values nearby.
• CT data types include measurement data, raw data, convolved data, and reconstructed data.

Basic Principles of CT (Computed Tomography)

• Algorithms are used to reconstruct images.
• Fourier transform describes the amplitude and phase of each sinusoid.
• Convolution is an image processing technique using a filter function.
• Interpolation estimates a function's value from surrounding known values.

Image Quality

• Spatial resolution is the ability to distinguish fine details.
• Noise is unwanted variations in image brightness.
• Contrast resolution is the ability to distinguish between tissues having subtle differences in density.
• Linearity is the output's consistency with input signal variations.
• Uniformity refers to the consistency of image properties.
• Artifacts include beam hardening, photon starvation, partial volume artifacts, out-of-field artifacts, and ring artifacts.
• Cone beam CT (CBCT) uses the operational physical principles of conventional CT, and has distinct instrumentation.
• Quality Control for CT scanners includes tests for proper operation and functionality.

Image Reconstruction from Projections

• Radon (1917) developed mathematical solutions for image reconstruction from projections.
• Early CT images were noisy because of reconstruction techniques used (Hounsfield).
• Convolution back-projection algorithms were introduced afterward improving image quality.
• CT aims to calculate µ-values (attenuation coefficients) from projection data using algorithms like back-projection, iterative methods, and analytic methods.

Reconstruction Algorithms

• Back projection (summation method or linear superposition) is a simple method but doesn't produce sharp images, often showing star-shaped patterns.
• Iterative algorithms assume initial values and repeatedly adjust these values until they match measured results. Methods include simultaneous iterative reconstruction, and iterative least-squares techniques.
• Analytic algorithms aim to address the issues of back-projection and iterative algorithms.
• Filtered back-projection filters out star-like blurring present in simple back projection images.
• Fourier reconstruction algorithms use frequency domain manipulations; to enhance image qualities.

Types of Data

• Measurement data: Subject to preprocessing to correct measurement errors before reconstruction.
• Raw data: Result of preprocessed scan data, used for image reconstruction steps.
• Convolved data: Applying a convolution filter to raw data for improved image quality.
• Image data: Result of reconstructing raw or convolved data, representing the final CT image.

3D Algorithms

• 3D reconstruction techniques for surface display involve preprocessing, segmentation, surface formation, and projection.

Image Manipulation and Windowing

• Image postprocessing modifies reconstructed images for improved viewing/interpretation.
• Windowing (gray-level mapping) modifies image contrast and brightness.
• Window width defines the range of CT numbers (gray levels) to display.
• Window level represents the center of a defined CT number range, and controlling display brightness.

Multiplanar Reconstruction (MPR)

• Reformats images into different planes (coronal, sagittal, paraxial) from a stack of axial images, an image reformatting technique.
• MPR Advantages: Enables visualization of specific structures; Localization of lesions/fractures and bone fragments.

Image Quality

• Reconstruction methods utilize raw data, while reformation methods use pre-existing image data
• Advantages and Disadvantages of MPR.