CT Physics, Radiation Dose & Hounsfield Units

Forming the image. A CT scanner rotates an X-ray tube and opposing detectors around the patient and measures X-ray attenuation from many angles. A computer reconstructs those profiles, by filtered back-projection or iterative reconstruction, into cross-sectional images of tissue attenuation. Modern helical multidetector CT acquires large volumes rapidly and allows multiplanar and 3D reconstruction, the basis of orthopaedic 3D planning.

The Hounsfield scale. Each voxel is assigned a Hounsfield unit, or CT number: its attenuation normalised to a scale on which water is 0 HU and air is -1000 HU. Know where water, air and bone sit and the rest of the scale can be interpreted from them.

View larger

View larger

Windowing. The full Hounsfield range cannot be displayed at once, so a chosen range is mapped to the grey scale. The window level sets the central HU and the window width the range spread across the grey scale. A narrow soft-tissue window and a wide bone window therefore display different tissues from the same data, which is why a fracture is seen best on a bone window.

View larger

View larger

The acquisition parameters. kVp, the tube voltage, sets the beam energy and penetration. mAs, tube current multiplied by time, sets the number of photons and so the image noise. Pitch (table travel per rotation), slice thickness and field of view govern coverage, resolution and noise, and every one of these settings trades image quality against radiation dose.

Measuring the dose. Three metrics, each built on the one before:

• CTDIvol (mGy) is the average dose within the scanned volume, measured in a standard phantom (16 cm head, 32 cm body). It reflects dose intensity per rotation, is independent of scan length, and is a scanner-output index rather than the patient's true dose.
• DLP (mGy·cm), the dose-length product, is CTDIvol × scan length, the total energy imparted over the whole scan.
• Effective dose (mSv) is DLP × a region-specific conversion factor (k). It weights for tissue radiosensitivity, so that stochastic (cancer) risk can be compared across body regions and modalities.

The magnitudes. A chest radiograph is around 0.02 mSv, a head CT about 2 mSv and an abdomen/pelvis CT roughly 5-10 mSv, against natural background radiation of about 2-3 mSv per year. An abdominopelvic CT is therefore equivalent to several years of background, the concrete basis for justification and ALARA.

Which way each parameter moves the dose. Change one setting and the dose moves predictably:

• mAs. Dose is directly proportional, so halving the tube current-time product halves the dose and raises the noise.
• kVp. Dose rises steeply and faster than in proportion, because kVp increases both the number and the energy of the photons. Lowering kVp in a small patient or a child saves a disproportionate amount.
• Pitch. At a fixed tube current, increasing pitch spreads the same output over more table travel and reduces dose (and z-axis resolution). Many scanners hold quality constant by automatically raising mAs as pitch rises, so pitch must always be read alongside the displayed CTDIvol rather than on its own.
• Slice thickness. Thinner slices improve resolution but need more photons per slice to hold noise constant, so they cost dose.
• Scan length. The parameter that costs nothing to get right: DLP grows with every centimetre covered, so trimming the coverage is the cheapest dose reduction available.

ALARA. CT delivers substantially more dose than plain radiography. The ALARA principle (As Low As Reasonably Achievable) requires the lowest dose that still answers the clinical question, supported by automatic exposure control, iterative reconstruction, appropriate kVp selection and a field of view limited to the region of interest. It matters most in young patients and women of childbearing age, in whom the lifetime risk of radiation-induced malignancy is greater.

Harmonised protocols. CT protocols vary widely and are often incomplete, which leads to inconsistency and unnecessary dose. Total hip arthroplasty planning shows it: across 17 published protocols the recommended kV (100-150) and mAs (100-250) varied widely, and most protocols were incomplete. What is needed is a harmonised low-dose protocol, optimised for both the field of view and the acquisition parameters, that balances diagnostic quality against patient dose.

The artefacts. Metal is the orthopaedic problem, but it is one of five:

• Beam hardening. The polychromatic beam loses its low-energy photons passing through tissue, producing cupping and dark streaks between dense structures.
• Metal/streak. Dense implants cause beam hardening, photon starvation and scatter, and the bright and dark streaks obscure peri-prosthetic bone.
• Partial volume. A voxel spanning two tissues is averaged, which misleads at boundaries and is worse with thick slices.
• Motion. Blur and ghosting.
• Ring. A faulty detector element.

Reducing dose and artefact. Four techniques lower the dose, the artefact or both:

• Iterative (and deep-learning) reconstruction models the acquisition physics and statistics to give lower noise at a given dose, allowing roughly a third-to-half dose reduction compared with filtered back-projection.
• Automatic tube-current modulation varies mAs with patient attenuation, so dose is delivered only where it is needed.
• Lower kVp suits smaller patients and iodine-based angiography.
• Dual-energy CT enables material decomposition and high-keV monoenergetic and metal-artefact-reduction reconstructions that specifically reduce the orthopaedic metal artefact.

View larger

View larger

View larger

Lead shielding can raise the dose. In a phantom study of CT pulmonary angiography in pregnancy, abdominal shielding increased the mother's effective dose by 47% and raised the mean fetal absorbed dose from 0.10 to 0.12 mGy instead of reducing it. The shield sits within the field seen by automatic exposure control, which responds by driving the tube current up. Shortening the scan to the anatomy actually needed cut fetal dose by 70%, from 0.10 to 0.03 mGy.

View larger

View larger

The fetal dose in proportion. The fetal dose measured from a standard CT pulmonary angiogram was around 0.1 mGy, three orders of magnitude below the 50 to 60 mGy organ doses associated with the tripled childhood cancer risks above. A CT of a limb or shoulder in a pregnant patient is further still from the uterus. The judgement is about justification and scan length; refusing an indicated scan is not the answer.

Fluoroscopy is a different problem. The shielding finding applies to CT with automatic exposure control, where a lead blanket inside the scanned field misleads the scanner into raising output. It does not transfer to fluoroscopy in theatre, where the operator and staff stand in scattered radiation and personal lead remains the mainstay. Occupational protection, dose limits and the pregnant-staff pathway are covered under radiation safety in orthopaedics.

The clinical indications for CT, its orthopaedic applications and a fuller treatment of pitch and reconstruction sit in CT imaging principles, and the metal-artefact problem specifically in MRI metal artefact reduction.