Radiation therapy image-matching glossary
The vocabulary a radiation therapy student meets at the treatment console, in plain English. Each entry says what the term means, why it matters, and — where there is one — what you actually see on screen when you practise it.
Educational use only. These are definitions of the field's own vocabulary for learners. They are general background, not clinical advice or protocol — always follow your department's procedures.
#
4DCT
A 4DCT sorts the acquired projections into respiratory phases, giving a set of CT volumes across the breathing cycle instead of one blurred average. It is how a moving target’s excursion is measured. The lung cases here are built from a 4DCT: the volume you register is the end-exhale phase of a real research dataset.
A
Adaptive radiotherapy
Adaptive radiotherapy changes the plan during the course when the anatomy no longer matches the one it was made on — weight loss, a shrinking tumour, a cavity that has filled. Daily imaging is what detects it: a shift that keeps growing in the same direction, or a target that no longer sits inside its margin, is a trigger to re-scan rather than to keep correcting.
Anatomical directions
Anterior is toward the front and posterior toward the back; superior (cranial) is toward the head and inferior (caudal) toward the feet; medial is toward the midline and lateral away from it. Left and right always mean the PATIENT’S left and right, not yours looking at the image — that single convention is behind most direction mistakes in setup.
Anatomical planes
Axial (or transverse) slices cut across the patient like a loaf; coronal splits front from back; sagittal splits left from right. A CT volume can be resliced into any of them, which is what the three CBCT panes are — the same voxels viewed three ways, so a shift you make in one appears immediately in the other two.
B
Beam's eye view (BEV)
The beam’s eye view is the view from the source looking down the beam axis — what the field "sees". It is the natural frame for shaping an aperture, because the target’s outline in that view is exactly what the leaves have to conform to. The trainer draws its apertures in that view, at the isocentre plane.
Bolus
Bolus is tissue-equivalent material laid on the skin so that the beam’s build-up region falls inside it rather than inside the patient, bringing the surface up to full effect. It is used where disease reaches the skin. It is not modelled in this trainer’s cases.
Boost
A boost is a further phase that treats a reduced volume — usually the original gross disease — after a larger volume has been treated. It concentrates the remaining treatment where the risk is highest while sparing what surrounds it. It is a planning and prescribing decision; this trainer works one fraction at a time and does not represent phases.
C
Cone-beam CT (CBCT)
A cone-beam CT is reconstructed from projections taken by the on-board kV imager as the gantry rotates around the patient in the treatment position. It is noisier and has more scatter than a planning CT, but it is acquired on the machine, at the moment of treatment.
Couch axes
Lateral is patient left–right, longitudinal is head–foot, and vertical is up–down. The trainer reports them as a couch correction, so the sign convention is the one printed beside the readout — Match review states the direction in plain words for the match in front of you.
Couch kick
A couch kick is a rotation of the couch about the vertical axis, so the beam no longer lies in the plane the gantry sweeps — a non-coplanar arrangement. It buys entry angles that a coplanar plan cannot reach, and it is the setup with the highest collision risk, because the patient is now turned into the gantry’s path. The SRS case makes you certify that path before it will move.
Craniospinal irradiation (CSI)
Craniospinal irradiation treats the whole brain and the spinal canal together. The target is far longer than one field can cover, so it is treated from several isocentres along the patient with junctions between them — which is what makes setup accuracy matter more than usual, because an error at a junction becomes a dose gap or an overlap rather than a small miss. The Full Treatment CSI workflow here images three stations 20 cm apart, has you record each one and enter the averages, then applies one common correction and reimages before treating each site in turn.
CT acquisition parameters
Slice thickness sets how finely the volume is sampled along the patient — thin slices reformat and reslice cleanly, thick ones look stepped in coronal and sagittal views. The reconstruction kernel trades sharpness against noise (a bone kernel is crisp and grainy, a soft-tissue kernel smooth). Field of view, kV and mAs set what is included and how noisy it is. The CT Simulation workspace exposes these with display-only consequences; no dose is calculated.
D
Deep-inspiration breath-hold (DIBH)
Deep-inspiration breath-hold has the patient hold a full breath so the chest wall lifts away from the heart, and treats only inside that held window. The trainer’s breath-hold case is voice-coached: you talk the patient into the gate and hold them there through each exposure.
Dose–volume histogram (DVH)
A dose–volume histogram summarises a plan by showing, for each structure, how much of its volume receives at least a given amount — the curve planners read to compare one plan with another. It is a planning-system output and depends on a dose calculation. Nothing in this trainer calculates dose, so no DVH exists here; what is graded is geometric setup accuracy.
DRR
A DRR — digitally reconstructed radiograph — is an X-ray-like image computed from the planning CT by summing attenuation along each ray. It is the reference the daily setup image is matched against, so it shows the patient exactly as planned rather than as positioned today.
E
Electron cone
Electrons scatter strongly in air, so an electron field is collimated by an applicator — a cone that carries the collimation down close to the skin, ending a few centimetres above the patient. That proximity is the point and also the hazard: with a cone fitted, the clearance to the patient drops sharply, which the trainer’s QA session has you feel directly.
F
Fiducial marker
Fiducials are small inert markers — usually gold — implanted in or near the target so it can be seen directly on an image that would otherwise show only bone. They matter when the target moves independently of the skeleton, which is why the prostate cases use them.
Field junction
Where two fields meet, small setup errors turn into a dose gap or overlap, because the edges are divergent and steep. The answers are geometric: block one half of each field so the edges are vertical, calculate the gap from the depth and field sizes, or move the junction between fractions (feathering) so no single plane of tissue sits at the seam all course. The CSI workflow here treats three isocentres with deliberate overlap regions for that reason.
Flattening filter and FFF
The flattening filter is a cone of metal in the head that evens out an intrinsically peaked beam so the profile is flat across the field. Removing it — flattening-filter-free — gives a much higher output rate with a peaked profile, which suits small stereotactic fields where flatness matters less than speed. Energies here are labelled X for photons and E for electrons, with FFF marked.
Fraction
A fraction is one treatment session — a course is divided into a number of them, delivered over days or weeks. Splitting treatment up lets normal tissue recover between sessions, and it is also why setup accuracy matters so much: the same position has to be found again every single time. Full Treatment in this trainer walks one fraction end to end.
G
Gantry
The gantry is the rotating arm carrying the treatment head; it turns about a horizontal axis through the isocentre, so the beam can enter from any angle around the patient. Angles are read in the machine’s own convention — the trainer’s gantry runs a signed range with real hard stops, which is why it sometimes has to unwind the long way round rather than take the short arc.
Graticule
The graticule is the crosshair and centimetre scale drawn over an image to mark the isocentre and give distance a reference. It is an overlay, not anatomy — in this trainer it is drawn at the isocentre plane, so a tick is a true centimetre there and not at the detector.
GTV, CTV and PTV
GTV is the disease that can be seen, CTV adds the microscopic spread around it, and PTV adds a geometric margin for setup and motion uncertainty. Contours drawn in this trainer are teaching geometry — several are explicitly synthetic expansions, not clinician-delineated volumes.
H
Hounsfield units
Hounsfield units are the CT density scale: water is 0, air about −1000, fat slightly negative, soft tissue a little above zero, and cortical bone from several hundred upward. Window and level select which part of that range is mapped to the visible greys, which is why a bone window and a soft-tissue window of the same slice look so different.
I
IGRT
IGRT — image-guided radiation therapy — means imaging the patient in the treatment position, comparing that image with the plan, and correcting the position before the beam is on. Everything this trainer simulates is the imaging-and-correction half of that loop.
Image artefacts
An artefact is structure in the image that is not structure in the patient. Dense metal starves the detector and throws bright and dark streaks; movement during acquisition smears edges; a faulty detector element draws a ring; scatter and beam hardening leave the middle of a large patient looking falsely dark (cupping). Cone-beam images carry more scatter than a planning CT, so they look flatter and noisier — that is the modality, not a fault.
Image quality
Spatial resolution is how fine a detail can be separated; contrast resolution is how small a density difference can be seen; noise is the random speckle that hides both. They trade against each other and against imaging dose — a lower-dose acquisition is noisier. It is why bone matching survives a poor image and soft-tissue matching does not.
Image registration
Registration is finding the transform that brings two image sets into the same frame of reference. What you do in this trainer is a rigid registration performed by hand: you translate and rotate the daily image onto the reference until the anatomy agrees, and the transform you have applied is read back as the couch correction.
Imaging protocol
How often a patient is imaged, and what is done with the result, is protocol rather than physics. Online correction means imaging, matching and moving the couch before this treatment; offline protocols image the first few fractions, average the errors and correct the systematic part for the rest of the course. Tight-margin work images daily and online; conventional sites may not.
Immobilisation
Immobilisation is the hardware that makes a position reproducible: it holds the patient the same way every day and limits how far they can move during treatment. It is what converts a good plan into a deliverable one, because a margin is only as small as the setup is repeatable. The trainer renders the real device for each case and includes it in collision checking.
Indexing
Indexing means locking the device into a numbered slot on the couch top instead of sliding it anywhere: the same board at the same index sits at the same longitudinal position every day, so the recorded couch values mean something. The trainer models the lattice as stations H5 through F9 at a 140 mm pitch, and moving a board one station changes the planned longitudinal value by exactly 14.0 cm.
Interfraction and intrafraction motion
Interfraction motion is change between one treatment and the next — different bladder filling, weight change, a slightly different setup. Intrafraction motion happens during a single treatment: breathing, peristalsis, the patient settling. Imaging before the beam catches the first; only imaging or gating during delivery catches the second.
Interlock
An interlock is a condition that must be satisfied before the beam or a motion is permitted — the right accessory fitted, the imager stowed, the door closed. A deadman is the complement: a control you must keep holding, so motion stops the moment you let go. The trainer uses both, which is why couch and gantry moves run only while the control is held.
Isocentre
The isocentre is the fixed point in the room where the beam axes cross — the gantry, collimator and couch all rotate about it. Planning puts that point at a chosen place inside the patient, and setup is the job of putting it back there each day. Every reticle and graticule in the trainer is centred on it.
K
kV vs MV imaging
kV imaging uses a separate diagnostic-energy X-ray tube, so it gives good bone and soft-tissue contrast and is what the on-board imager and cone-beam CT use. MV imaging uses the treatment beam, which gives a much flatter, lower-contrast image but proves where the field actually lands.
L
Linear accelerator (linac)
A linear accelerator accelerates electrons down a waveguide to megavoltage energies. They either strike a target to produce an X-ray beam, or are spread out and used directly as an electron beam. Everything else on the machine — gantry, collimator, couch, imagers — exists to point that beam at the right place. The trainer’s machine is a generic teaching model, not any manufacturer’s design.
M
Machine QA
Machine QA proves the linac still does what it is trusted to do, and each test has its own frequency rather than the session having one. This trainer runs a single QA session covering three protocol blocks: imaging localisation on a phantom, output constancy on a detector array, and an MLC picket fence — each block one room setup. It is a teaching exercise in the workflow and the judgement, not a measurement of a real machine, and it is deliberately not graded.
Monitor unit (MU)
A monitor unit is the machine’s own unit of beam output, counted by the ionisation chambers in the treatment head as the beam runs. The linac is calibrated so a set number of MU delivers a defined amount under reference conditions — so MU is a machine quantity, not a patient dose. Every MU figure in this trainer is fictional, and nothing here calculates dose.
Motion management
A target that moves with breathing can be treated through the motion (an ITV drawn to cover its whole excursion), have the motion reduced (abdominal compression), or be treated only in part of the cycle (gating or breath-hold). Gated and breath-hold techniques watch a surrogate — a marker block, the surface, a spirometer — because the tumour itself is not visible in real time.
MR imaging
MR uses magnetic fields rather than X-rays, so it separates soft tissues that look almost identical on CT — which is why a cranial target is usually outlined on MR even though the plan is calculated on CT. It carries no Hounsfield scale, because its signal is tissue behaviour rather than density. Two of the volumetric cases here are MR: you register them exactly as you would a cone-beam, in the same three planes with the same six axes, but the window and level read as relative brightness rather than HU.
Multileaf collimator (MLC)
The multileaf collimator is a bank of opposed tungsten leaves in the treatment head that shapes the field; the jaws are the four larger blocks behind them. The trainer models 60 opposing pairs and four independent jaws, with the aperture drawn at the isocentre plane.
O
Off-bone matching
An off-bone case is one where the target has moved relative to the skeleton — bladder and rectal filling carry the prostate, for example. A perfect bony match then leaves the target displaced, so you register the target or its fiducials instead and accept that the bones will not line up.
On-board imager (OBI)
The on-board imager is the kV X-ray tube and flat panel carried on retractable arms, mounted at right angles to the treatment beam. It takes the planar setup images and, by rotating around the patient, the cone-beam CT. In the trainer you deploy and retract those arms yourself, and a deployed arm is part of the collision model — extend the panel before an AP exposure, and stow it before the gantry swings.
Organ at risk (OAR)
An organ at risk is a normal structure close enough to the target that its exposure has to be limited — cord, heart, lung, rectum, parotid, depending on the site. Contours for them are drawn at planning and shown on the daily image so you can see what a shift moves toward. Several structures in this trainer are automated or explicitly synthetic teaching contours rather than clinician-drawn ones, and are labelled as such.
Output constancy
Output constancy asks whether the beam still delivers what it was calibrated to deliver, so that a monitor unit means the same thing today as on the day the machine was commissioned. A detector array is set up at a known distance and each energy is delivered to it in turn. In this trainer the electron energies run first so the applicator is mounted once, then the photon energies; every reading is a labelled scenario value, not a dose calculation. Machine output is not a console adjustment — an out-of-criterion result is a physics finding.
P
Patient position (HFS, FFS, supine, prone)
Patient position is recorded as the direction of entry plus the posture: HFS is head-first supine, FFS feet-first supine, and the prone variants HFP and FFP have the patient face-down. It matters because it decides how patient directions map onto machine axes — the same couch move is superior for a head-first patient and inferior for a feet-first one. Nearly every case here is head-first supine, and the trainer applies that mapping for you.
Patient preparation
Pelvic setup depends on organ filling, so preparation is part of the treatment: a comfortably full bladder pushes bowel out of the field and a consistently empty rectum keeps the prostate where it was planned. It is never perfect, which is exactly why the prostate cases here are off-bone — the gland and its seeds move relative to the pelvis as filling changes.
Penumbra
The field edge is not a line: the source has a finite size, so the dose falls off over a few millimetres — the penumbra. Leaves add their own effects, transmitting a little through the leaf body, leaking between neighbours, and under-dosing along the stepped tongue-and-groove sides. The picket-fence QA session renders all three as visible features of the film.
Picket fence test
A picket fence checks that every MLC leaf arrives where it was told. The leaves stop at a set of evenly spaced positions during one delivery, so the film shows a row of stripes; a leaf out of calibration bends its stripe. In this trainer the couch is driven fully out, the EPID is the detector, and one moving arc is delivered and then analysed twice — a qualitative read of the strips, and a quantitative per-bank one against an absolute reference. A whole-bank shift moves the fitted line with it, which is why the second read exists.
Pitch, roll and yaw
Pitch is nodding about the left–right axis, roll is tipping about the head–foot axis, and yaw is turning about the vertical axis. Because an AP and a lateral image both look along axes that yaw preserves, yaw is the one rotation a planar pair cannot resolve.
Planning CT
The planning CT is the scan the whole plan is built on: acquired in the treatment position, with the immobilisation in place and reference marks applied, so that everything measured from it can be reproduced on the machine. Every DRR you match against is computed from it, which is why it counts as the patient "as planned".
Portal image
A portal image is taken with the treatment beam itself, on the flat panel opposite the head (the EPID). Because it uses the megavoltage beam it has poor soft-tissue contrast, so it is read for bone and for the field edge — it shows what the beam actually sees.
Q
QA phantom
A phantom is a manufactured object that stands in for a patient, so a machine can be tested against geometry that is known exactly. This trainer carries two: a cube with five embedded markers — one at its centre plus four on the body diagonals — used for imaging localisation, and a ball phantom used for Winston-Lutz. Both are simulator-owned teaching geometry rather than any commercial product, and the cube is hollow on purpose: a solid block of that size would swamp the markers it exists to show.
R
Radiation safety
Staff protection rests on time, distance and shielding: nobody is in the room during delivery, the vault walls and maze absorb what escapes, and door interlocks drop the beam if that is breached. An area monitor shows the room state and personal dosimeters record what each person actually received. ALARA is the governing principle — as low as reasonably achievable, not merely under the limit.
Region of interest
A registration is only as good as what you let it look at. Clinical systems restrict the match to a region — a box around the vertebrae, the prostate, the seeds — so that irrelevant anatomy cannot pull the result; automatic algorithms then optimise a similarity measure such as mutual information over the grey values inside it. Bone matching is robust and reproducible; soft-tissue matching is what you need when the target moves independently of bone.
Rigid vs deformable registration
A rigid registration allows only translation and rotation — the anatomy keeps its shape, which is what a couch can actually reproduce. Deformable registration warps locally to account for anatomy that has genuinely changed shape. Daily setup correction is rigid, and so is every match this trainer grades.
Room lasers
The room lasers project fixed planes that intersect at the isocentre — two lateral, one sagittal overhead. They are the only visible reference for where the isocentre is with the patient on the couch, so setup begins by matching the skin marks to them. In CT Simulation the movable green lasers do the mirror-image job: they mark on the patient where the planned isocentre falls.
S
SAD, SSD and SID
SAD is source-to-axis distance, fixed at 100 cm on a treatment machine — the source sits one metre from the isocentre. SSD is source-to-surface distance, measured to the skin where the beam enters, so it changes with the patient and the angle. SID and SDD are source-to-imager distances for the MV and kV panels, and both are adjustable — moving a panel changes magnification and clearance at once.
SBRT and SRS
SBRT and SRS deliver a large dose in very few fractions to a small, well-defined target. The tight margins leave almost no room for setup error, which is why the stereotactic cases here carry the tightest tolerances — the intracranial SRS case grades to 1 mm and 1°.
Setup error
Setup error is the difference between where the patient is and where the plan says they should be. A systematic error repeats every fraction — often something baked in at simulation — and so shifts the whole course; a random error varies day to day and averages out somewhat. Systematic errors carry the greater weight in margin formulas, which is why they are worth catching early.
Setup tattoos
Setup tattoos are permanent ink dots placed at simulation where the room lasers crossed the skin. They are the starting position: the therapists line the tattoos to the lasers to get close, and then imaging refines it. They mark a point on the skin, not the target — which is exactly why imaging is still needed after them.
Six degrees of freedom (6DOF)
Six degrees of freedom are the three translations (lateral, longitudinal, vertical) plus the three rotations (pitch, roll, yaw) a couch can correct. A planar AP-and-lateral pair cannot resolve yaw, so 2D/2D corrects five; a volumetric registration resolves all six.
T
Thermoplastic mask
A thermoplastic mask is a perforated sheet warmed until pliable, moulded over the patient and clipped to a baseplate; it hardens to hold the head — or head, neck and shoulders — in one place. Three-point masks cover the head, five-point ones extend over the shoulders. Cranial and head-and-neck setups use them because the tolerances there are the tightest on the machine.
Time-out and identification
The defence against treating the wrong person, or the right person wrongly, is procedural: identify the patient by two independent identifiers before anything else, confirm the site and the plan, and let the record-and-verify system refuse any setting that does not match the approved plan. Imaging is one layer of that chain, not a substitute for it.
Topogram (scout)
A topogram — also called a scout — is the single flat projection a CT takes by driving the couch through a stationary tube. It is not diagnostic; it exists so the range of the real scan can be set against visible anatomy. The CT Simulation workspace acquires an AP and a lateral topogram and has you drag the superior and inferior limits on it.
Treatment boards
Boards mount on the couch and set the pose: a breast board inclines the torso and takes the arms overhead, a wing board gives an arms-up thoracic or abdominal setup a T-grip to hold, and a head-and-neck baseplate carries the headrest and mask clips. The trainer mounts the right board per case and solves the couch height from where the patient’s own scanned surface rests on it.
Treatment couch
The couch is the patient support, and the only thing that actually corrects a setup error: the correction you dial is a couch move. A modern couch offers six degrees of freedom — vertical, longitudinal and lateral translation plus rotation, pitch and roll. Readouts are absolute encoder values on the machine’s own scale, which is why the trainer shows both the absolute position and the signed shift you are applying.
Treatment field
A field is one beam: its gantry, collimator and couch angles, its jaw settings and leaf pattern, and how much output it delivers. A plan is a set of fields chosen so that their overlap covers the target while each passes through different normal tissue. A static field is delivered from one fixed angle; an arc delivers while the gantry moves.
Treatment head
The treatment head holds the beam line: the target that converts electrons to X-rays, the primary collimator, the flattening filter or its absence, the monitor chambers that measure output, and then the jaws and multileaf collimator that shape the field. Its front face is the nearest part of the machine to the patient, which is what makes it the usual collision risk.
Treatment plan
The plan is everything decided before the first treatment: the target and normal-structure contours, the isocentre, the beam or arc arrangement, the shaping, and the intended output. Setup imaging exists to reproduce its geometry. The plans in this trainer are teaching geometry — they are not optimised, no dose is calculated, and none is deliverable.
Triggered imaging
Triggered imaging takes images during delivery — at set gantry angles, time intervals or monitor units — to confirm the target has not moved mid-treatment. The prostate Full Treatment fraction models it: a triggered kV outside the gate holds the beam, and you re-image, re-match and drive the couch before delivery resumes.
V
Vacuum bag and cradle
A vacuum bag holds beads that lock rigid when the air is drawn out, and an expanding-foam cradle is poured around the patient in a tray and sets. Both take the shape of that one patient, so the body sits back into the same hollow each day. The CT Simulation workspace casts these from the patient’s own scanned surface.
View names (AP, PA, lateral)
A view is named for the direction the beam travels through the patient: AP enters the front and exits the back, PA the reverse, and a lateral passes side to side. An orthogonal pair — an AP or PA plus a lateral — is the classic setup pair, because two views at right angles pin down all three translations. What they cannot pin down is rotation about the axis they share, which is why 2D/2D here corrects five degrees of freedom rather than six.
VMAT
VMAT delivers while the gantry rotates, with the leaves and dose rate changing continuously through the arc; IMRT shapes the beam from a set of fixed angles. The Full Treatment arcs animate that motion — they are a teaching visualisation, not a calculated or deliverable plan.
W
Waveguide and accelerator
An electron gun injects electrons into an accelerating waveguide, driven by microwave power from a klystron or magnetron; a bending magnet turns the beam onto the patient axis, and it strikes a target to produce X-rays. Monitor ionisation chambers just below then measure every fraction of the output — they are what monitor units are counted by. The Machine Room draws all of these as labelled components; the internal geometry is representative teaching geometry, not manufacturer design.
Wedge
A wedge is an attenuator, thick on one side and thin on the other, that progressively absorbs more of the beam across the field and so tilts the distribution. It compensates for a sloping surface or shapes the overlap of two beams. Modern machines usually achieve the same effect by sweeping a jaw during delivery instead of inserting metal.
Winston-Lutz test
A Winston-Lutz test checks that the radiation isocentre and the imaging isocentre are the same point — the assumption every image-guided correction rests on. A small ball is set at isocentre and imaged at several gantry angles; the ball’s offset from the field centre on each film gives the error in three dimensions. The Machine Room carries a ball phantom for this: place a deliberate offset, image at orthogonal angles, read the millimetre offset each film reports, correct with the couch and re-image. A couch shift cannot fix a genuine imaging-versus-radiation disagreement — that is a machine finding, not a setup one.
Practise the vocabularyMatch real CT and CBCT data against planning references, with live residual-error feedback.
Open the trainer