WHAT IT IS

A CT scanner (computed tomography, also called CAT scan) is an X-ray machine that takes hundreds to thousands of projection images of a body from many angles and uses a computer to reconstruct them into cross-sectional images (slices) and 3D volumes. The word comes from Greek tomos (slice) and graphein (to write). A plain X-ray flattens the whole body onto one picture, so a rib, a lung nodule, and the spine all overlap. CT removes the overlap: each voxel (3D pixel) in the output is a number that says how strongly that tiny block of tissue weakens the X-ray beam.

The machine is a large ring called the gantry with a bore (opening) usually 70 to 90 cm across, and a motorized patient table that slides the patient through it. Inside the gantry, an X-ray tube and a curved array of detectors sit on opposite sides of the bore and spin together around the patient. A complete modern head or chest exam is a few seconds of actual X-ray time. Most hospitals of any size have at least one, and trauma centers and emergency departments depend on them for fast diagnosis.

CT is a structural imaging method. It shows density and shape: bone, air, fat, fluid, calcium, blood, contrast-filled vessels, tumors. It does not directly show function in the way PET does, and it has less soft-tissue contrast than MRI, but it is fast, widely available, tolerant of patients who cannot lie still for long, and works with metal implants that rule out many MRI exams.

HISTORY

Mathematical roots. In 1917 the Austrian mathematician Johann Radon proved that a function in two or three dimensions can be recovered from the integrals of that function along lines through it (the Radon transform). That is the math behind CT, though it had no practical use in 1917. The South African-born physicist Allan MacLeod Cormack independently worked out the problem for X-ray attenuation in the early 1960s and published his results in 1963 and 1964. William Oldendorf did a laboratory demonstration of reconstruction in 1961 using a rotating model.

Hounsfield and EMI. The English electrical engineer Godfrey Newbold Hounsfield, working at EMI's Central Research Laboratories in Hayes (the same company known for The Beatles' records), built the first working clinical CT scanner. His early lab prototype used a gamma source and took days to acquire data; an X-ray version cut that to hours and then minutes. The first patient scan was done in October 1971 at Atkinson Morley's Hospital in Wimbledon, London; it showed a cystic frontal-lobe brain lesion. Hounsfield and the radiologist James Ambrose presented the results in April 1972, and the scanner was announced publicly later that year. The EMI Mark I was head-only. Its image was an 80 × 80 matrix, and one slice took about 5 minutes to scan, followed by long computer reconstruction. Hounsfield and Cormack shared the 1979 Nobel Prize in Physiology or Medicine.

Whole-body CT. The first whole-body scanners appeared in 1974-1976, including the ACTA scanner by Robert Ledley at Georgetown University and the EMI CT5000 whole-body unit. From there the field moved fast, and EMI's lead evaporated once GE, Siemens, Philips and Toshiba entered. EMI left the CT business by the early 1980s.

GENERATIONS OF SCANNER GEOMETRY

• First generation (1971): a single pencil beam and a single detector that translate across the head and then rotate 1 degree, repeated 180 times. Minutes per slice.
• Second generation (mid 1970s): a narrow fan beam with a handful of detectors, still translate-rotate but with larger rotation steps. Tens of seconds per slice.
• Third generation (late 1970s to present): a wide fan beam covering the whole patient and a detector arc of several hundred elements rotating together with the tube. No translation. Rotation time fell to a few seconds, then to well under one second. Every modern scanner is third generation.
• Fourth generation: a fixed full ring of detectors with only the tube rotating. It was used by some manufacturers but the cost of the full ring and scatter issues meant it was largely abandoned for medical use.
• Electron-beam CT (EBCT, 1980s): no moving tube; an electron beam is steered magnetically around a stationary tungsten target ring. Used for fast cardiac imaging (calcium scoring). Made obsolete by fast rotating gantries.
• Slip ring and helical (spiral) CT (about 1989-1990): slip rings replaced the power and data cables, so the gantry can rotate continuously in one direction. With the table moving at the same time, the tube traces a spiral around the patient and acquires a whole volume in a single breath-hold. Willi Kalender and Siemens are usually credited with the first clinical spiral CT around 1989.
• Multi-slice (multi-detector row, MDCT): an early two-row system (Elscint Twin) appeared in 1992, and 4-slice scanners from several vendors in 1998. 16-slice came in 2001-2002, 64-slice around 2004, and the Toshiba (now Canon) Aquilion ONE with 320 rows and 16 cm of z-coverage in 2007 could image a whole heart or brain in one rotation.
• Dual-source CT (2005-2006): two tube/detector systems mounted about 90 degrees apart on one gantry, introduced by Siemens (SOMATOM Definition). Temporal resolution is a quarter of the rotation time, around 83 ms on the first model and about 75 ms on the Definition Flash (2008) and about 66 ms on the later Force and NAEOTOM Alpha.
• Dual-energy and spectral CT: imaging at two X-ray spectra to separate materials (iodine, calcium, uric acid, water). Done with two tubes (Siemens), fast kVp switching (GE), or a dual-layer detector (Philips IQon, 2013).
• Photon-counting CT (clinical since 2021): detectors that count each X-ray photon and measure its energy directly instead of converting to light first. The Siemens Healthineers NAEOTOM Alpha received US FDA clearance in September 2021 and was the first clinical scanner of the type.

HOW IT WORKS: THE CHAIN

In order, a scan runs through this chain: tube generates X-rays, filter and collimator shape the beam, the beam passes through the patient, detectors measure what gets through, the data acquisition system digitizes it, a slip ring carries the data off the spinning gantry, and a reconstruction computer turns it into images.

THE X-RAY TUBE

The tube is a vacuum glass or metal envelope with a heated tungsten filament (cathode) that boils off electrons, and a rotating anode target that the electrons are accelerated into by a high voltage. Typical CT tube voltages are 70 to 150 kVp (kilovolts peak), with 80, 100, 120 and 140 the standard choices; 120 kVp is the common default for adult body imaging. Tube current is typically tens to several hundred milliamperes, and the highest-end tubes go beyond 1000 mA for large patients and fast scans. Peak generator power is often in the range of roughly 60 to 120 kW for modern units.

Less than 1 percent of the electron beam energy becomes X-rays; the rest becomes heat in a focal spot only a millimeter or so across. To survive, the anode is a disk of tungsten-rhenium alloy on a molybdenum or graphite body (graphite stores heat) spinning at roughly 7,000 to 10,000 rpm (about 150 Hz) so the focal track is constantly renewed. The key CT tube spec is anode heat storage, measured in MHU (million heat units); typical values on current tubes are about 5 to 8 MHU, with anode cooling rate in kHU per minute being equally important because it sets how long a series of scans can continue without a cooldown wait. Some designs (Siemens Straton and later) are rotating-envelope tubes where the whole tube spins and the anode is cooled directly by oil in contact with its back; others use liquid-metal (gallium-alloy) bearings and a conventional anode. These cope with the mechanical load of the gantry.

The gantry's rotation puts enormous force on the tube. At roughly 0.25 to 0.28 seconds per revolution and a tube radius of about 0.6 m, centripetal acceleration works out to the order of 30 g (approximate, calculated from those numbers). That is why tubes and detector modules are built as heavy, balanced, rigid assemblies, and why bearing failure and imbalance are real service concerns.

FILTRATION AND COLLIMATION

The tube produces a spectrum from low to high energy photons. The low-energy photons would be absorbed by the first few centimeters of the patient and add dose without contributing to the image, so the beam is filtered with aluminum, copper or other metal plates (total filtration at least about 2.5 mm aluminum equivalent in regulation). A bow-tie filter, a shaped piece thicker at the edges than at the center, flattens the beam intensity after it passes through a round patient and further reduces dose at the skin. Siemens introduced a selectable tin pre-filter on some systems to narrow the spectrum for low-dose lung and sinus protocols. Collimators (metal shutters) define the width of the beam along the body axis, which sets the slice thickness and the coverage and keeps scatter and dose down. A second set of collimating grids on the detector faces rejects scattered photons.

DETECTORS

Almost all clinical CT scanners built until recently use indirect-conversion (scintillator) detectors: X-rays hit a scintillator crystal or ceramic, commonly a rare-earth ceramic such as gadolinium oxysulfide (GOS) or a garnet (GE Gemstone), cesium iodide in some designs, that glows; a photodiode behind it turns the light into a current. The detector arc has several hundred to well over 800 channels across the fan and 16 to 320 rows along the body axis, in modular cassettes. Indirect detection loses some signal in the light conversion and the septa between pixels.

Direct-conversion detectors use a semiconductor such as cadmium telluride (CdTe) or cadmium zinc telluride (CZT), in which each X-ray photon directly creates a cloud of electron-hole pairs that a high bias voltage sweeps to pixel electrodes. In a photon-counting detector the pulse height is proportional to photon energy, so each photon is counted and sorted into energy bins. Benefits: no electronic noise floor in the counting, no septa needed and so smaller pixels and sharper images (the NAEOTOM Alpha offers an ultra-high-resolution mode with details to about 0.2 mm), inherent spectral information on every scan, and better contrast-to-noise at low dose. The difficulties are pile-up at high flux, crystal defects and charge sharing, and expensive crystals, so these detectors came to market decades after they were first proposed. Philips offers an alternative, the dual-layer "sandwich" detector (a top layer that absorbs low-energy photons and a bottom layer that gets the rest), which is still a scintillator system.

DATA ACQUISITION AND SLIP RINGS

Each detector channel's analog signal is integrated over short intervals (thousands of views per rotation, typically around 1,000 or more), digitized by the data acquisition system (DAS) mounted on the gantry, and sent off the spinning part as a high-speed digital stream (multiple gigabits per second on modern units) via the slip ring or an optical or capacitive link. The slip ring is a set of conductive rings with brushes (or contactless couplings) that pass power up to the rotating assembly and carry data back. Replacing cables with slip rings is what made continuous rotation and helical CT possible.

PATIENT TABLE

The table is a carbon-fiber top (strong, light, and nearly transparent to X-rays) on a motorized base. It moves in and out of the gantry with sub-millimeter positioning accuracy and carries patients up to roughly 200 kg or more depending on model (bariatric tables go up to about 300 kg on some systems, per manufacturer specs). In helical mode the table speed is synchronized with the gantry. The table can also raise and lower, and the gantry can tilt on some scanners.

PHYSICS: HOW AN IMAGE IS BUILT

Attenuation (Beer-Lambert law). A monochromatic beam of intensity I₀ passing through a uniform material of thickness x and linear attenuation coefficient μ comes out with intensity I = I₀ e^-μx. For a non-uniform body, the measurement along one ray gives the line integral of μ, i.e. ln(I₀/I) = ∫ μ(s) ds. Each detector reading in each view is one such line integral (a projection value). The collection of all of them over all angles is a sinogram, and reconstruction means recovering μ(x,y) from it. Real X-ray beams are polychromatic, which causes the artifacts discussed below.

Hounsfield units. The reconstructed attenuation values are scaled so that water is 0 and air is -1000: HU = 1000 × (μ - μ_water) / (μ_water - μ_air). Typical values at about 120 kVp (approximate and they vary by scanner, kVp and patient):

• Air: about -1000 HU
• Lung tissue (aerated): about -900 to -500 HU
• Fat: about -120 to -80 HU (commonly quoted near -100)
• Water: 0 HU; cerebrospinal fluid and simple cysts: about 0 to 15 HU
• Urine and bile: about 0 to 20 HU (varies)
• Brain gray matter: about 35 to 45 HU; white matter: about 25 to 35 HU
• Unclotted blood: about 30 to 45 HU; acute clotted blood (fresh hemorrhage): about 50 to 80 HU
• Muscle: about 40 to 60 HU; liver (unenhanced): about 50 to 70 HU
• Contrast-filled vessel (arterial phase): about 200 to 500 HU or higher
• Kidney stones and other calcifications: about 200 to over 1000 HU, depending on composition (uric acid stones are lower, roughly 200 to 450 HU)
• Cancellous (spongy) bone: about 200 to 400 HU; dense cortical bone: about 500 to over 1500 HU
• Metal: saturates at the top of the scale (3071 HU on typical 12-bit systems)

Reconstruction. The classic method is filtered back projection (FBP): each projection is convolved with a ramp-like filter (the kernel) to compensate for the blurring that simple back projection causes, and the filtered projections are smeared back across the image at their original angle and summed. FBP is fast and was the clinical standard for decades. Its weakness is noise at low dose. Since about 2008-2010 vendors introduced iterative reconstruction (GE ASiR and Veo, Siemens IRIS, SAFIRE and ADMIRE, Philips iDose and IMR, Canon AIDR 3D and FIRST). These loop between an image estimate and a model of the scanner's physics and noise, repeatedly refining the image; they allow roughly 30 to 60 percent or more dose reduction at similar image noise depending on the algorithm and task. Since about 2018-2020 deep-learning reconstruction has arrived (GE TrueFidelity, Canon AiCE, Siemens' neural-network-based tools, Philips Precise Image), in which a neural network trained on high-dose images is used to denoise or reconstruct low-dose data. For cone-beam geometry, the FDK (Feldkamp-Davis-Kress) algorithm is the standard approximation.

Windowing. The monitor cannot display the whole -1000 to +3000 HU range in 256 gray levels, so the radiologist picks a window level (center) and window width. A brain window is something like level 40, width 80; a lung window about level -600, width 1500; a bone window about level 400, width 1800 or more; an abdomen soft-tissue window about level 40, width 400. The same data reveal totally different things in different windows.

Pitch, slice thickness and resolution. In helical scanning, pitch is the table travel per gantry rotation divided by the total beam collimation width. A pitch of 1 means contiguous coverage; below 1 overlaps (more dose, better sampling), above 1 gaps (faster, less dose, interpolation). Dual-source scanners in a special fast mode can use pitch up to about 3.2 to 3.4 (Siemens specs for Flash and Force) to freeze the heart or image a child without sedation. Slice thickness is set by the detector row width and the reconstruction; modern scanners acquire submillimeter data (commonly 0.5 to 0.625 mm) and reconstruct thicker slices for routine reading. Spatial resolution in the plane is typically about 0.3 to 0.5 mm in the best conventional systems and about 0.2 mm in photon-counting ultra-high-resolution modes, and nearly isotropic in the z-axis. Temporal resolution is about half the rotation time in single-source scanners (around 135 to 140 ms for 0.27 to 0.28 s rotation), and about a quarter in dual-source (66 to 75 ms). That is the difference between a blurred and a sharp coronary artery. Low-contrast resolution (the ability to see small differences in HU against noise) is limited by dose and noise, and is why CT is weaker than MRI for soft tissue.

Gantry rotation speed. Fastest commercial rotation times are about 0.25 s (Siemens Force and NAEOTOM Alpha), 0.27 s (Canon Aquilion ONE Prism, approximately) and 0.28 s (GE Revolution series). Mechanical limits and the g-force on the tube set the floor.

CONTRAST AGENTS

Natural CT contrast between soft tissues is small, so most body CT is done with iodinated contrast injected into a vein (usually through a power injector at roughly 2 to 5 mL per second, with total volumes around 50 to 150 mL, concentrations of roughly 300 to 370 mg iodine per mL). Iodine has a high atomic number and a K-edge at 33.2 keV, so it absorbs diagnostic X-rays strongly; a vessel or highly vascular tumor lights up white. Current agents are non-ionic, low- or iso-osmolar: iohexol (Omnipaque), iopamidol (Isovue), iodixanol (Visipaque), iopromide (Ultravist), ioversol (Optiray). Timing matters: arterial phase for angiography, portal-venous phase (about 60 to 70 s) for most abdominal work, delayed phase for the urinary tract or washout. Bolus tracking watches an artery's HU and triggers the scan when it hits a threshold. Oral contrast (dilute iodine or barium) is sometimes given to light up the bowel, and rectal or intra-articular contrast is used for specific studies. Non-contrast CT is the default for stroke bleeds, kidney stones and many lung and bone questions.

Risks. (1) Allergic-like reactions: mild ones (hives, itching, nausea) occur in roughly 0.5 to 3 percent of injections with modern low-osmolar agents (estimates vary), and severe reactions are rare (on the order of 0.01 to 0.04 percent). Prior reaction raises risk; premedication protocols exist. It is not a true iodine or seafood allergy. (2) Kidney effects: contrast-associated acute kidney injury is uncommon in patients with normal kidney function; current ACR and National Kidney Foundation consensus (2020) puts the real risk mainly in patients with eGFR under 30 mL/min/1.73 m² or acute kidney injury, and screening focuses on those patients. Hydration is the main preventive measure. (3) Extravasation, when contrast leaks out of the vein into the surrounding tissue, which is usually minor but can cause tissue injury in large volumes. (4) Interaction with metformin, which guidelines handle by kidney function. (5) Thyroid effects in susceptible patients (iodine load). Pregnancy and lactation have their own guidance. A technologist or radiologist screens every patient for these.

RADIATION DOSE

CT is a high-dose X-ray modality compared with plain films, and it accounts for a large share of the medical radiation received by the population in developed countries. Doses are quoted as CTDIvol (mGy, scanner output), DLP (dose-length product, mGy·cm) and an effective dose in millisievert (mSv) estimated as DLP times a body-region conversion factor (roughly 0.002 for head, 0.014 for chest, 0.015 for abdomen/pelvis mSv per mGy·cm). Effective dose is a population-risk tool, not a precise individual measurement. Typical adult values (approximate and highly dependent on scanner, protocol and patient size):

• Chest X-ray (PA): about 0.02 to 0.1 mSv
• Head CT: about 1 to 2 mSv
• Low-dose lung cancer screening CT: about 1 to 1.5 mSv
• Chest CT: about 5 to 7 mSv (often less on a modern low-dose protocol)
• Abdomen and pelvis CT: about 8 to 10 mSv
• Coronary CT angiography: about 1 to 5 mSv on modern scanners (older retrospective-gated: 10 to 20 mSv or more)
• For comparison: natural background radiation averages about 2.4 mSv per year worldwide (UNSCEAR) and about 3 mSv per year in the US from natural sources; a transatlantic flight is about 0.05 to 0.1 mSv

The principle is ALARA (as low as reasonably achievable): scan only when the clinical question justifies it, and use the lowest dose that answers it. Scanners reduce dose with automatic exposure control (tube current modulation by patient size along the z-axis and by angle during the rotation), ECG-based dose modulation in cardiac CT, prospective ECG triggering, lower kVp for small patients and contrast studies, tin filtration, iterative and AI reconstruction, organ-dose modulation (reduced exposure over the breasts, eyes), and size-specific protocols for children (Image Gently campaign). The cancer risk from a single CT is small and difficult to measure directly; epidemiological estimates are mostly extrapolated from atomic-bomb survivor data using a linear no-threshold model, which is debated. The risk is higher in children because of longer remaining lifespan and more sensitive tissue.

WHAT CT CAN DETECT

• Head and stroke: non-contrast CT is the first test for suspected stroke because it quickly shows hemorrhage (bright, about 50 to 80 HU acutely) and rules it in or out before clot-busting therapy. Early ischemic changes (loss of gray-white differentiation, hyperdense vessel sign) are subtle. CT angiography finds large-vessel occlusions for thrombectomy, and CT perfusion estimates salvageable brain.
• Trauma: whole-body (pan-scan) CT in major trauma finds head bleeds, spinal fractures, aortic injury, solid organ lacerations (liver, spleen, kidney), pneumothorax and free blood in the abdomen within minutes.
• Pulmonary embolism: CT pulmonary angiography shows clots as filling defects in contrast-filled pulmonary arteries.
• Aortic disease: aneurysm, dissection and rupture on CT angiography.
• Cancer: detection, staging (size, local invasion, lymph nodes, metastases to lung, liver, bone), biopsy guidance, planning of radiation therapy, and follow-up of response. Low-dose CT screening of high-risk smokers reduces lung cancer mortality (US National Lung Screening Trial, about 20 percent reduction).
• Lung: nodules, pneumonia, emphysema, interstitial disease (high-resolution CT), and COVID-19 pneumonia patterns (ground-glass opacities).
• Heart: coronary artery calcium scoring (Agatston score, a non-contrast scan) predicts cardiac risk; coronary CT angiography shows plaque and narrowing noninvasively with a high negative predictive value for ruling out significant disease; also structural heart planning (e.g. before valve replacement).
• Abdomen: appendicitis, diverticulitis, bowel obstruction, pancreatitis, abscess, liver and kidney masses, and inflammatory bowel disease complications.
• Urinary tract: kidney and ureter stones on low-dose non-contrast CT, which is very sensitive for stones; CT urography for tumors.
• Bone and joints: complex fractures (pelvis, spine, wrist, ankle, skull base), surgical planning, and 3D reconstructions for orthopedic and craniofacial surgery.
• Sinuses, ears and teeth: sinus disease, temporal bone detail, and dental/jaw imaging (often cone-beam CT).
• Infection and abscess anywhere, and guided drainage.
• Virtual colonoscopy (CT colonography) and virtual bronchoscopy, built from CT data.
• Interventional guidance: CT-guided needle biopsy, ablation and drainage.
• Forensics: post-mortem CT (virtopsy) for fractures, gas and foreign bodies, and unknown remains.

What CT is poor at: subtle soft-tissue differences (small brain tumors, spinal cord, ligaments, early ischemia), small bowel and pelvic organ detail, and function (metabolism) which needs PET or SPECT. PET/CT and SPECT/CT combine a CT scanner with a nuclear camera so the CT supplies the anatomy and attenuation correction.

CT VERSUS MRI, X-RAY AND ULTRASOUND

• X-ray (radiography): cheapest and fastest, very low dose, good for bones and chest, but a 2D shadow. CT gives 3D detail at 50 to 100 times the dose of a chest film.
• CT: fast (seconds), widely available, good for bone, lung, bleeding, vessels (with contrast), and acute emergencies; uses ionizing radiation; moderate soft-tissue contrast; works with pacemakers (with precautions) and metal implants better than MRI does. Cost is moderate.
• MRI: no ionizing radiation; far better soft-tissue contrast (brain, spinal cord, joints, ligaments, pelvis, liver); long exams (20 to 60+ minutes), loud, needs the patient to hold still, and has contraindications for some implants and devices and for ferromagnetic metal. Poor for bone detail and lung. Cost is higher; machines use superconducting magnets with liquid helium.
• Ultrasound: no radiation, portable, real-time, cheap, ideal for pregnancy, heart (echo), gallbladder, thyroid and vessels; limited by bone and air, and operator-dependent.
• PET and SPECT: nuclear functional imaging, usually combined with CT (PET/CT) for anatomical localization.

LIMITS AND ARTIFACTS

• Beam hardening: a polychromatic beam loses its soft low-energy photons first, so the average energy rises as it passes through the body. The result is cupping (the center of a uniform object looks darker) and dark streaks between dense objects such as the petrous bones in the skull base. Corrected with filtration, calibration and software; dual-energy synthesized monoenergetic images reduce it.
• Metal artifacts: dental fillings, hip prostheses, spinal screws, pacemaker leads and surgical clips absorb nearly all photons (photon starvation) and produce bright and dark streaks. Remedies include higher kVp, thin slices, metal artifact reduction algorithms (MAR, O-MAR, iMAR) and high-energy virtual monoenergetic images.
• Motion: breathing, heartbeat, bowel movement and patient movement blur and double structures. Faster rotation, breath-holds, ECG gating and dual-source scanners reduce it.
• Partial volume effect: when a voxel contains more than one tissue it shows an average HU, so tiny structures can be hidden or small calcifications under-measured. Thinner slices reduce this.
• Ring artifacts: concentric rings from a miscalibrated or failing detector channel (a common sign a detector module needs service).
• Cone-beam and helical artifacts: windmill and streak patterns from wide detector rows in the z-direction; corrected by reconstruction algorithms.
• Truncation: if the patient is wider than the scan field of view, the edges of the image are corrupted.
• Noise: especially in larger patients and low-dose scans, trading off against dose.
• Incidentalomas and overuse: CT often finds harmless abnormalities that lead to follow-up tests; a healthcare issue, not a technical limit.

CONE-BEAM CT, INDUSTRIAL AND SECURITY CT

Cone-beam CT (CBCT) uses a flat-panel detector (a 2D array, as in digital X-ray) and a cone-shaped beam, and acquires a whole volume in a single rotation of the source around the patient or object, with no table movement. It is used in dentistry and oral surgery (compact scanners with a seated or standing patient), ENT, C-arm angiography suites (intraoperative 3D imaging), and radiation therapy linacs (image-guided radiotherapy, IGRT, for positioning the patient before treatment). Its image quality for soft tissue is lower than a multi-slice CT, and Hounsfield values are less reliable because of scatter, but the machines are smaller, cheaper and lower dose for the specific tasks.

Industrial CT scans manufactured parts to find internal voids, cracks, porosity in castings, weld flaws and assembly errors, and to measure internal geometry (metrology) without cutting. Micro-CT reaches micron-scale resolution for electronics, small biological specimens, fossils and materials research. Industrial systems often use the specimen rotating in a fixed beam, with tube voltages from tens of kV up to 450 kV or more, and linear accelerators of several MeV for large castings and rocket motors.

Security CT is used to screen luggage. Explosive-detection systems (EDS) in airports use CT to compute the density and effective atomic number of objects in bags and flag possible explosives, and in recent years the TSA and European airports have rolled out CT scanners at passenger checkpoints so that laptops and liquids may stay in bags (details vary by airport and rules change over time). Vendors include Smiths Detection, Rapiscan (OSI), Analogic and Leidos.

POWER, INSTALLATION AND ROOM

The figures below are typical and approximate. The manufacturer's site planning guide and a qualified medical physicist control the actual requirements for any given scanner.

• Power: three-phase supply, commonly 380 to 480 V (depending on region and model), with a demand of very roughly 50 to 150 kVA peak for high-power units. The tube generator draws large short bursts, so a stiff supply (low line impedance) matters. A UPS is normally used for the computer and console, not the generator and tube.
• Room size and weight: a CT suite is usually a scan room of roughly 25 to 40 m² plus a control room and an equipment/technical area. A gantry weighs on the order of 1.5 to 3 tonnes (depending on model), so the floor loading and the route into the building (door width, elevator, sometimes a crane through a wall opening) must be planned.
• Radiation shielding: walls, doors, windows and sometimes floor and ceiling are lined with lead (commonly on the order of 1.5 to 2.5 mm of lead equivalent, but this is calculated for each room by a physicist using workload and occupancy per national guidance such as NCRP Report 147 in the US), with a lead-glass control window, interlocked warning lights and door signage. Scatter from the patient is the main source for shielding; the primary beam is intercepted by the detector.
• Cooling and HVAC: the gantry, generator and computers dump kilowatts of heat. Scan rooms are usually held at roughly 18 to 24 °C with humidity controlled to roughly 30 to 60 percent non-condensing (manufacturer specific). Many systems use a water-glycol chiller for the tube and generator, or air/oil heat exchangers, in addition to room air conditioning. A few minutes of tube cooling between heavy series is normal.
• Other: a quality-assurance phantom for daily air calibration and weekly checks, a contrast power injector, image archive (PACS) network connection, and emergency stop and patient intercom.

MAINTENANCE AND FAILURE POINTS

• X-ray tube: the number one consumable and the biggest service cost (often tens of thousands of dollars installed, depending on model). Typical life is highly variable and workload-dependent, commonly quoted at around 1 to 3 years, or on the order of 100,000 to 300,000 scan-seconds / 100,000 to 200,000+ rotations or exposures on some tubes (rough, vendor-dependent figures). End of life shows as arcing (sparking), reduced output, filament failure, rising anode drag and bearing noise, and gas or vacuum loss.
• Tube warm-up and calibration: a cold tube must be gradually warmed up before high-power scans to avoid thermal shock to the anode. Daily air calibrations (a scan with no patient) and periodic detector calibrations keep CT numbers right.
• Detector modules: bad channels cause ring artifacts and require recalibration or module replacement; scintillator and photodiode degradation is slow.
• Slip ring and brushes: brush wear and contamination cause noise, data dropouts and power faults; contactless designs reduce this.
• Gantry bearings and drive belt: wear, noise and vibration; the bearing is a major mechanical item.
• Cooling system: pumps, fans, coolant level and filters; overheating shuts the scanner down.
• High-voltage generator and cables: HV faults, capacitors, and connectors.
• Computer, software and storage: disk and reconstruction hardware failures, software licensing, and cybersecurity (the console is a Windows or Linux computer on a hospital network and needs patching).
• Routine QA: daily phantom checks of CT number accuracy (water at 0 HU), noise and uniformity; annual physics testing for dose, resolution, slice thickness and alignment, per regulation and accreditation (ACR in the US).
• Service life: a scanner commonly lasts about 8 to 12 years in service; refurbished units are widely resold, often with a replacement tube.

MANUFACTURERS AND CURRENT MODELS

The market for hospital-grade CT is dominated by a handful of firms. Model names change, so treat the lists below as examples and check the maker for current line-ups.

• Siemens Healthineers (Germany): SOMATOM family. Dual-source: SOMATOM Definition Flash (2008), SOMATOM Force (2013), and the SOMATOM Drive; single-source: SOMATOM go. series (go.Up, go.Top, go.All), Edge Plus; and the photon-counting NAEOTOM Alpha family (Alpha, Alpha.Pro, Alpha.Prime, Alpha.Peak), the first clinical photon-counting scanner.
• GE HealthCare (USA): Revolution CT, Revolution Apex, Revolution Apex Elite, Revolution Maxima, Revolution Frontier, Revolution Vibe, Optima CT series; Gemstone garnet detector, TrueFidelity deep-learning reconstruction, and a deep-silicon photon-counting prototype under development (check current status).
• Philips (Netherlands): IQon Spectral CT (2013, first dual-layer detector clinical scanner), Spectral CT 7500, CT 5300, CT 3500, Incisive CT, and Philips Big Bore.
• Canon Medical Systems (Japan; formerly Toshiba Medical): Aquilion ONE (320-row, 16 cm), Aquilion ONE Prism and Genesis editions, Aquilion Prime SP, Aquilion Lightning, Aquilion Serve, and the Aquilion ONE INSIGHT; with the AiCE deep-learning reconstruction. A photon-counting CT is in research and development at Canon.
• United Imaging (China): uCT 530, uCT 760, uCT 780, uCT 960+, uCT Atlas; a growing global presence.
• Others: Fujifilm (acquired Hitachi's diagnostic imaging business in 2021), Neusoft (China), Samsung/NeuroLogica (portable head/body CT, e.g. BodyTom, OmniTom), and others in smaller segments; specialty makers include Carestream, Planmeca and Dentsply Sirona (dental CBCT), and Varian/Elekta (CBCT on linacs).

Key performance classes (approximate): entry scanners 16 to 64 slices at 0.4 to 0.5 s rotation; mainstream 64 to 128 slices at 0.33 to 0.35 s; premium 128 to 320 rows at 0.25 to 0.28 s; dual-source at 0.25 s with 66 to 75 ms temporal resolution; photon-counting at 0.25 s with 66 ms and ultra-high-resolution mode. The scanner in the first photo is a Siemens SOMATOM Definition Flash-class dual-source system and the second shows a NAEOTOM Alpha photon-counting scanner.

SAFETY NOTE

This page is a technical reference, not medical advice. Whether a CT scan is needed for a given person is a decision for their doctor, who weighs the diagnostic benefit against the dose.

IN ONE SENTENCE

A CT scanner spins an X-ray tube and a detector arc around the patient, measures how much each ray is attenuated from thousands of angles, and reconstructs those line integrals into a 3D map of Hounsfield-unit densities, in seconds, at a dose of roughly 1 to 10 mSv: the fastest, most widely used way to look inside the body for bleeds, clots, stones, fractures and tumors.