Filtered back projection
FBP is an analytical reconstruction method. It is computationally efficient and historically established, but image noise and streak artefact become increasingly conspicuous as photon statistics deteriorate.
CT revision
CT for FRCR Part 1 Physics.
CT accounts for six of the forty questions in the current RCR Physics distribution. The useful revision target is not a list of scanner facts: it is being able to predict what happens to image quality, artefact and radiation output when an acquisition or reconstruction parameter changes.
Attenuation and Hounsfield units
- CT reconstructs a map related to the linear attenuation coefficient of each voxel. Attenuation depends on photon energy and tissue composition, including density and effective atomic number.
- The Hounsfield scale normalises attenuation to water: water is approximately 0 HU and air approximately −1000 HU. Materials that attenuate more strongly than water have positive values. The familiar scale is useful, but measured HU is not a fixed intrinsic constant for every material: it can vary with beam energy, scanner, reconstruction and artefact.
- Photoelectric absorption is strongly dependent on atomic number and photon energy, while Compton scatter is more closely related to electron density over the diagnostic range. Their changing relative contributions help explain both image contrast and energy-dependent attenuation.
Helical acquisition and pitch
- In multidetector helical CT, pitch is commonly defined as table travel per gantry rotation divided by the total nominal collimated beam width. A higher pitch therefore advances the patient further for each rotation and covers a given z-axis length more quickly.
- Pitch changes the sampling geometry and can affect longitudinal resolution, interpolation and artefact. Very high pitch can reduce z-axis sampling density, although modern reconstruction methods and scanner designs modify the practical effect.
- Do not memorise “higher pitch always means lower dose” as a universal rule. With fixed tube current and otherwise comparable acquisition, increasing pitch reduces exposure per unit length; however, modern automatic exposure-control systems may change tube current to maintain a target image quality. The actual dose relationship therefore depends on scanner mode and exposure-control behaviour.
Reconstruction and image texture
Iterative reconstruction
Iterative methods repeatedly compare an estimated image or projection set with the measured data and update the estimate. They can suppress noise and some artefacts and permit different dose–image-quality trade-offs, but the appearance and texture of the reconstructed image can change. A particular percentage dose reduction is not universal across scanners, algorithms or clinical tasks.
Resolution, noise and section thickness
- Spatial resolution describes the ability to separate small objects or high-spatial-frequency detail. In-plane resolution is influenced by detector sampling, focal spot, field of view, matrix and reconstruction kernel; longitudinal resolution also depends on acquisition and reconstructed section geometry.
- Contrast resolution is the ability to distinguish tissues with small attenuation differences and is strongly limited by image noise. Quantum noise falls as the number of detected photons rises.
- Reconstructing thinner sections reduces partial-volume averaging and improves z-axis localisation, but each image voxel contains fewer photons and is noisier if the acquisition is otherwise unchanged. Thicker sections improve SNR at the cost of more partial-volume averaging.
- Slice thickness and reconstruction interval are not the same. The interval controls the spacing between reconstructed image centres. Overlapping reconstructions can improve multiplanar reformats and reduce stair-step appearance, but they do not manufacture spatial resolution that was not present in the source acquisition.
- A sharper reconstruction kernel preserves more high-frequency detail but also makes noise more conspicuous. A smoother kernel suppresses noise at the expense of edge sharpness.
Tube output and automatic exposure control
- mA × time = mAs. With other factors fixed, increasing mAs increases the number of emitted photons. More detected photons reduce relative quantum noise, while radiation output rises approximately in proportion to mAs.
- kVp changes the X-ray spectrum as well as output. Increasing kVp increases photon energy and penetration and also changes subject contrast. Its effect on dose is not a simple linear analogue of mAs.
- Automatic exposure control changes tube current according to estimated attenuation, often longitudinally and/or through the gantry rotation, to achieve a system-specific image-quality target. Correct patient positioning matters because scout geometry and apparent patient size can influence modulation.
- When answering parameter questions, state what is being held constant. A relationship observed at fixed mA may not remain true when the scanner automatically compensates for a parameter change.
CT dose quantities: know what they are not
| Quantity | What it represents | Common trap |
|---|---|---|
| CTDIvol (mGy) | A standardised scanner-output index derived from measurements in reference CT phantoms and adjusted for pitch in helical imaging. | It is not the absorbed dose to an individual patient's organs. |
| DLP (mGy·cm) | CTDIvol multiplied by the irradiated scan length, giving a practical index of total scanner output over that examination range. | It is not a direct measurement of patient effective dose. |
DLP can be combined with region- and population-specific conversion coefficients to estimate effective dose for broad radiation-protection comparisons. Effective dose is itself a population-level protection quantity, not a precise prediction of one patient's absorbed organ doses or individual risk.
Artefacts and their physical causes
| Artefact | Mechanism / appearance | Mitigation concept |
|---|---|---|
| Beam hardening | Cupping or dark bands/streaks, particularly between dense structures, as lower-energy photons are preferentially removed from a polychromatic beam. | Beam-hardening correction, appropriate filtration/calibration, higher tube potential when clinically justified, and projection-based or iterative correction methods. |
| Partial volume | A voxel containing several tissues is assigned an averaged attenuation value, obscuring small structures or creating misleading intermediate density. | Use thinner reconstructed sections or smaller voxels when adequate source data and noise permit. |
| Motion | Blurring, misregistration or streaking from patient, respiratory or cardiac motion. | Shorter acquisition, appropriate breath-hold instruction, immobilisation and gating where clinically appropriate. |
| Metal | Severe streaks and dark/bright bands caused by photon starvation, beam hardening, scatter and incomplete projection data around very dense material. | Metal-artefact-reduction reconstruction, iterative methods, suitable acquisition parameters and, on dual-energy systems, appropriately selected virtual monoenergetic reconstructions. |
| Photon starvation | Streaking through very attenuating paths when too few photons reach the detector for stable reconstruction. | Automatic tube-current modulation, adequate exposure for patient size and suitable reconstruction/noise-reduction methods. |
Dual-energy CT: the useful core
- Acquiring data with sufficiently different X-ray spectra provides information about how attenuation changes with energy. Materials that overlap on a conventional CT number can therefore sometimes be separated using their energy-dependent behaviour.
- Depending on the platform, spectral information can support material-decomposition images such as iodine maps, virtual non-contrast reconstructions and virtual monoenergetic images. These are mathematical reconstructions derived from the spectral data rather than literal additional acquisitions at every displayed energy.
- Virtual monoenergetic energy selection is a trade-off: lower energies can increase iodine contrast but may increase noise, whereas higher energies can reduce some beam-hardening and metal-related artefact at the expense of contrast. The optimal setting is task-dependent.
Test yourself
Use these as verbal concept checks. They are revision prompts, not copied examination or question-bank items.
- What does a Hounsfield unit compare with water, and why can the measured value vary with photon energy and reconstruction?
- How is helical pitch defined, and why is its effect on dose not safely described by one universal rule on modern scanners?
- What is gained and lost when reconstructed slice thickness is reduced?
- How does reconstruction interval differ from slice thickness, and why can overlapping images improve multiplanar display without creating new acquired resolution?
- Why does increasing mAs usually reduce quantum noise, and what role does kVp play in penetration and contrast?
- What do CTDIvol and DLP describe, and why is neither an individual patient's effective dose?
- How does automatic exposure control respond to changing patient attenuation?
- Which CT artefacts arise principally from beam spectrum, insufficient photons, motion, metal or voxel averaging?
- What extra information can two energy spectra provide compared with a conventional single-energy acquisition?
Test yourself
One FRCR Physics-style sample stem
This is newly authored public sample content and is not taken from the protected question bank. Mark each statement True or False, then check the explanations.
Regarding CT acquisition and image quality:
True
Thinner sections contain fewer detected photons per reconstructed voxel, increasing noise, while reducing averaging across different tissues.
False
CTDIvol is a standardised scanner-output index derived from reference phantoms, not an individual organ-dose measurement.
True
At fixed tube current, increasing pitch can reduce exposure per unit length, but automatic exposure-control systems may alter tube current or other parameters. The resulting dose effect is therefore scanner- and mode-dependent.
True
Sharper kernels improve edge detail at the cost of increased visible noise compared with smoother kernels.
True
Spectral information from different X-ray energy distributions can support material separation and derived reconstructions such as iodine maps.
Physics revision guides
Choose your next guide
Core candidate guides
Exam format and syllabus distribution
The 40-stem, 200-item paper, timing, marking, current topic distribution and official candidate guidance.
Pass mark and standard setting
How modified Angoff and Hofstee standard setting work, and why there is no permanent fixed percentage.
Revision strategy
A practical route from concept learning and topic blocks to mixed practice, error repair and full mocks.
True/False exam technique
Statement-level pacing, absolutes, units, proportionality, trade-offs and a disciplined review method.
Mock exams and question-bank practice
Use targeted stems and exam-length mocks for different jobs, then convert each result into focused recovery work.
Topic revision guides
Matter and Radiation
FRCR Physics matter and radiation revision: atomic structure, photon energy, attenuation, interactions, inverse square law and common True/False traps.
Radiography and Fluoroscopy
FRCR Physics radiography and fluoroscopy revision: X-ray production, detectors, grids, AEC, image quality, fluoroscopic dose and common exam traps.
Radionuclide Imaging
FRCR Physics radionuclide revision: decay, gamma cameras, collimators, SPECT, PET, counting statistics, corrections, dose and common True/False traps.
Radiation Safety
FRCR Physics radiation safety revision: dose quantities, biological effects, IRR17, IR(ME)R 2017, duty holders, optimisation, DRLs and common traps.
Ultrasound
FRCR Physics ultrasound revision: acoustics, transducers, pulse and beam properties, resolution, Doppler, artefacts, safety and common True/False traps.
Magnetic Resonance Imaging
Precession, relaxation, encoding, k-space, SNR trade-offs, artefacts and MRI safety.
Practise what you have revised
Work five complete stems with no account, or sit an exam-length mock with a free account.
Further reading
The Physics Bank is independent examination preparation. It is not affiliated with or endorsed by the Royal College of Radiologists. Exam rules change: always check the current RCR candidate guidance for your own sitting.