CT Physics, Radiation Dose & Hounsfield Units

A Hounsfield unit, or CT number, is a measure of the X-ray attenuation of a voxel, normalised to a scale on which water is defined as zero and air as minus one thousand. On that scale, fat is around minus one hundred, soft tissue is roughly plus thirty to sixty, acute blood a little higher, and dense cortical bone and metal are about plus one thousand and above. Because the whole range of Hounsfield values cannot be displayed at once, we use windowing: the window level sets the central Hounsfield value and the window width sets the range mapped across the grey scale, so a bone window and a soft-tissue window display different structures from the same dataset - which is why a subtle fracture is best appreciated on a bone window. The image itself is reconstructed from attenuation profiles acquired as the X-ray tube and detectors rotate around the patient. The key acquisition parameters then trade image quality against radiation dose: kVp is the tube voltage, which sets the beam energy and penetration; mAs is the tube current-time product, which determines the number of photons and therefore the image noise; and pitch and slice thickness affect coverage, resolution and noise. Increasing the mAs or using thinner slices reduces noise and improves resolution, but at the cost of a higher radiation dose. Dose is quantified by the CTDIvol in milligray, the dose-length product in milligray-centimetres, and the effective dose in millisieverts, and the ALARA principle requires using the lowest dose that answers the clinical question - aided by automatic exposure control, iterative reconstruction and low-dose protocols. This is especially important in young patients and women of childbearing age, in whom the radiation-induced cancer risk is greater.