CT Imaging Principles

Computed tomography removes the superimposition that limits plain radiography. Because every slice is a cross-section, fracture geometry, articular surface congruity, fragment displacement and the relationship of bone to the surrounding soft tissues can be seen directly in all three planes, and its spatial resolution for cortical bone (0.5-1mm) far exceeds MRI. When precise bony anatomy is the question, CT is the modality of choice.

What it is for. In orthopaedic practice CT characterises complex fractures, above all intra-articular ones, plans the operation, checks reduction and implant position afterwards, judges union and nonunion, stages tumours and evaluates spinal pathology. Acquisition takes seconds, a single dataset yields multiplanar reformats and 3D renderings, subtle non-displaced fractures missed by plain radiography are shown, and CT angiography provides vascular assessment in dislocations. Each of these is developed in the sections that follow.

What it costs. The major disadvantage is radiation dose, 2-20 mSv depending on the region, and it is worth being precise about the size of the difference because the two comparisons in common use do not agree. A CT of the pelvis (6-10 mSv) delivers roughly ten times the effective dose of an AP pelvic radiograph (around 0.7 mSv), but around 300-500 times that of a single chest radiograph (0.02 mSv). The very large multiples quoted in the literature are almost always against a chest film; against a radiographic series of the same region the factor is nearer 5-15. Both statements are true, and a candidate who names the comparator when quoting the number will not be caught out. Either way the absolute dose is high enough that clinical justification and dose optimisation matter for every CT requested.

The other limitations. Orthopaedic implants produce significant metal artefact through beam hardening and photon starvation. Soft tissue contrast is limited compared with MRI, so ligaments, cartilage and bone marrow oedema cannot be reliably evaluated. CT costs significantly more than plain radiography, it is not portable, so the patient must travel to the scanner, and iodinated contrast carries risks of allergy and nephrotoxicity.

Data Acquisition

The gantry. An X-ray tube is mounted on a rotating gantry opposite a detector array. The tube rotates continuously around the patient, 360 degrees in approximately 0.3-0.5 seconds on modern scanners, emitting a fan-shaped or cone-shaped beam; the tissues attenuate (absorb and scatter) the beam to varying degrees, and the detectors on the opposite side measure the intensity of the transmitted beam at thousands of angles around the patient.

Multi-detector CT. Modern scanners (MDCT) carry 64 to 320 detector rows and acquire multiple slices per rotation. That dramatically reduces scan time and gives isotropic voxel resolution, equal resolution in all three planes, which is what high-quality multiplanar reconstruction depends on.

View larger

Helical Acquisition and Pitch

Helical scanning. Modern CT is acquired helically (spiral): the table moves continuously while the tube rotates, so the X-ray focus traces a helix relative to the patient and a whole volume is acquired in one breath-hold or pass. The single most examined acquisition parameter is pitch:

Pitch = table travel per gantry rotation / total nominal beam width (collimation)

• Pitch = 1: the table advances exactly one beam-width per rotation (contiguous acquisition)
• Pitch over 1: the table moves faster than the beam width; faster coverage and (all else equal) lower dose, but interpolation gaps can reduce z-axis resolution and increase helical (windmill) artefact
• Pitch under 1: overlapping acquisition; higher z-axis resolution and better for fine detail, at the cost of higher dose

Choosing it. High pitch is chosen to freeze motion (polytrauma, paediatric or uncooperative patients) and low pitch for high-resolution work. Pitch is the trade-off between speed and coverage, dose, and z-axis resolution. Many scanners automatically adjust tube current with pitch to hold image quality constant, so always interpret pitch alongside the displayed CTDIvol rather than in isolation.

Hounsfield Units

The unit. The fundamental unit of CT imaging is the Hounsfield Unit (HU), named after Sir Godfrey Hounsfield who developed clinical CT. It is a linear transformation of the X-ray attenuation coefficient, normalised to water:

HU = 1000 × (μ tissue - μ water) / μ water

where μ is the linear attenuation coefficient. The normalisation gives water a value of exactly 0 HU and air approximately -1000 HU, a standardised scale across all CT scanners. Low values appear dark and high values white, which is why bone and metal are bright and fat and air are dark on the greyscale image.

View larger

The Hounsfield Unit as a Bone Density Measurement

The free measurement. Because the HU is a calibrated, absolute quantity rather than a relative greyscale, every CT that includes the trunk already contains a bone density measurement that nobody requested. A region of interest in the trabecular bone of the L1 vertebral body on any routine abdominal or lumbar CT gives a number that tracks DXA-defined osteoporosis closely: an L1 attenuation of 160 HU or less is around 90% sensitive, 110 HU or less is more than 90% specific, and the negative predictive value is 99% above 200 HU. This is opportunistic screening, and it costs nothing: no extra scan, no extra dose, no extra appointment.

Who it helps. The applications are immediate: a patient booked for spinal instrumentation, a periprosthetic fracture being planned for revision rather than fixation, or an elderly patient whose staging CT shows L1 at 90 HU and who therefore needs bone-health assessment before anything is screwed into their skeleton. It also catches the group DXA misses. In the source cohort, over half of the patients with a moderate or severe vertebral fracture had a non-osteoporotic T-score, and almost all of those had L1 attenuation of 145 HU or less.

Two caveats. Intravenous contrast raises measured attenuation, so thresholds derived from unenhanced scans cannot be applied directly to a portal-venous study, and values shift with kVp and with scanner calibration. Quote the threshold with the protocol it came from.

Window and Level Settings

Why windows exist. A CT dataset contains a far wider range of HU values than a monitor can display (typically 256 grey levels). Window width sets the range of HU values displayed, and window level (centre) sets the midpoint of that range.

Windowing changes the display, not the data. It selects which portion of the Hounsfield scale reaches the screen, nothing more. The same dataset can be viewed on bone windows to assess a fracture and on soft tissue windows to assess a haematoma without rescanning the patient.

Everything above, the Hounsfield scale, the windows, the reformats and the artefacts, exists to be applied in the same order on every scan. A fixed sequence is what stops the eye stopping at the first fracture it finds.

Both windows, every time. The common pitfall is to find the fracture on bone windows and miss the expanding haematoma, vascular injury or compartmental swelling that only the soft tissue windows show. In spinal trauma the canal is assessed on soft tissue windows for retropulsed fragments and epidural haematoma.

Choosing CT vs Other Modalities

A frequent viva theme is justifying CT over an alternative. The clinical question, whether it is bone geometry, soft tissue, marrow or vascular, should drive the choice of modality, not habit.

CT delivers substantially higher radiation doses than plain radiography and is the single largest contributor to medical radiation exposure in developed countries. Understanding CT dose metrics and optimisation strategies is essential for fellowship examinations.

Effective Dose by Region

The Quantities the Scanner Reports

A frequent physics-viva trap is to quote effective dose (mSv) without knowing the quantities the scanner actually reports. Distinguish three things:

Measured versus estimated. CTDIvol and DLP are physical dose quantities measured on a standard phantom; effective dose is a calculated, risk-weighted estimate, not a measured patient dose. The k-factor is larger for the trunk, with its radiosensitive viscera, and small for the extremities, which is why a limb CT carries a far lower effective dose than its CTDIvol alone might suggest. Diagnostic reference levels (DRLs) are benchmarked in CTDIvol and DLP, not mSv.

Dose Optimisation

On the scanner. Automatic tube current modulation adjusts the mAs to patient size and body region, iterative reconstruction takes the dose down further, low-kVp protocols suit the extremities, and the scan length is limited to the region of interest. Every scan needs an appropriate clinical indication and justification.

In the clinic. Plain radiography is the first-line investigation, and CT is reserved for when the films are insufficient for clinical decision-making. Avoid a repeat CT when previous imaging can answer the question, use MRI rather than CT for soft tissue questions (ligaments, bone marrow oedema), and consider low-dose CT protocols for follow-up imaging.

Dual-energy CT (DECT) acquires data at two different kVp settings simultaneously, so materials can be decomposed by their energy-dependent attenuation. Several applications matter in orthopaedics:

Gout in practice. DECT identifies monosodium urate deposits as small as 2mm, even without tophi on examination; commercial software colour-codes the crystals green and calcium purple/blue. Individual studies report sensitivity of 78-100% and specificity of 89-100%; the pooled meta-analysis in the Evidence Base gives 87% sensitivity and 84% specificity, and the pooled specificity sits below the range the individual studies report. False positives occur with nail bed artefact, skin calluses and motion artefact.

CT is the single largest contributor to medical radiation exposure worldwide. Across high-income health systems, CT accounts for only a minority of imaging studies (typically under 10% of examinations) yet contributes the majority (around 60-70%) of the collective dose from diagnostic radiology — the central justification for diagnostic reference levels (DRLs) and the ALARA principle. CT utilisation continues to rise globally, fastest in middle-income settings as scanner access expands, which makes appropriate-use criteria as relevant in resource-limited settings as in tertiary centres.

Global Epidemiology of CT Use

• CT volumes per capita are highest in the United States, Japan, and parts of Europe and lowest in low-resource settings where access — not over-use — is the dominant problem.
• Orthopaedic CT demand is driven by the ageing population, rising fragility-fracture burden, and the growing prevalence of in-situ metalwork (revision arthroplasty volumes are increasing across all major arthroplasty registries, including the NJR, AJRR, AOANJRR, and Scandinavian registries), which increases the need for metal-artefact-reduction CT.

Side-by-Side Society Guidance

High- vs Limited-Resource Practice Variation

• Well-resourced settings: routine thin-slice MDCT with iterative (or deep-learning) reconstruction, automatic tube-current modulation, dual-energy capability, and vendor MAR; CT is standard for preoperative planning of complex fractures and for 3D-printed patient-specific models.
• Limited-resource settings: older single- or few-slice scanners, limited dual-energy and MAR availability, and constrained access mean plain radiography (and, where available, MRI) carries more of the diagnostic load. Justification matters even more where each scan competes for scarce capacity, and protocol optimisation and shared reference levels are the highest-yield safety interventions.

Diagnostic reference levels, not billing pathways, are the appropriate global benchmark for governing CT use — dose audit against DRLs is recommended by ICRP, the IAEA, and national regulators regardless of health-system funding model.

CT physics is settled; what to do with the numbers is not. Four questions come up in vivas precisely because they have no agreed answer, and a candidate who names the reason each remains open is doing better than one who picks a side.

Whether an individual patient should be quoted a cancer risk at all

The convention is to apply the linear no-threshold (LNT) model, risk proportional to dose with no safe threshold, and multiply. The objection is that LNT was adopted as a radiation protection assumption, deliberately conservative, extrapolated downward from atomic-bomb survivor data at doses far above diagnostic range. Professional medical-physics bodies have argued that predicting the number of cancers caused by doses below about 50 mSv is not scientifically supportable, because the predicted excess is far smaller than the statistical noise around a lifetime cancer incidence of roughly 40%.

Why it stays unresolved. The effect, if it exists, is too small to measure. Detecting a fractional excess risk at 5 mSv against that background would require cohorts in the millions with complete confounder control. Large paediatric CT cohorts have reported dose-response relationships for leukaemia and brain tumours, but the standing objection is confounding by indication: children who get scanned differ from those who do not, sometimes because of the very condition that later declares itself. The pragmatic position for a viva is to give the number, name the model, and say plainly that it is a cautious upper bound rather than an observed effect. Neither "there is no risk" nor a confidently stated risk figure is defensible.

Effective dose is being used for a job it was not designed to do

Effective dose (mSv) is a population protection quantity: it is computed on reference phantoms using tissue weighting factors averaged across both sexes and all ages, for the purpose of comparing practices and setting limits. It is routinely quoted, including on this page, as though it were a dose delivered to the individual in front of you.

Why it stays unresolved. There is no equally portable alternative. Organ-dose-based, size-specific and age-specific risk calculators exist and are more defensible, but they require inputs that are not available at the point of the conversation, and clinicians will not use them. Effective dose survives because it is the only quantity that lets you compare a CT with a bone scan in one sentence. It is also why diagnostic reference levels are set in CTDIvol and DLP and never in mSv: the regulator is auditing a machine and a protocol rather than estimating a person's risk.

Immediate total-body CT versus selective imaging in trauma

Observational work suggested a survival advantage for scanning everyone; the randomised trial did not confirm it. In-hospital mortality was identical, and the additional radiation is not trivial.

Why it stays unresolved. The trial was conducted in systems where selective CT is fast and readily available, so it may have compared immediate total-body CT with something close to it. Time-to-diagnosis and missed-injury rates favour the pan-scan, and neither converts into a mortality signal in a population where most deaths follow from injuries that imaging cannot alter. The question has therefore shifted from "does it work?" to "in whom?", and no selection rule has yet been validated prospectively.

Numbers that are reproducible without being meaningful

Weight-bearing CT can measure middle facet subluxation, foot-ankle offset and syndesmotic volume with excellent inter-observer agreement. The same is true of CT-based union scoring. In both cases the measurement is far ahead of its interpretation: high reproducibility is not the same as a validated threshold for intervention.

Why it stays unresolved. Establishing that a measured value should change management requires longitudinal cohorts linking the number to an outcome, and those take years to mature; meanwhile the technology and the software measuring the number keep changing, which resets the clock. Treat these values as a way of describing a deformity or a gap precisely, not as a criterion for operating.