Clinical Oncology, Anatomy, and Radiobiology for the Dosimetrist
Domain 1 covers clinical oncology, cross-sectional anatomy and radiobiology as they appear at the planning console. The exam rarely asks you to diagnose; it asks whether you can turn a prescription, a stage and a set of organ-at-risk tolerances into a defensible plan, and whether you notice when a hypofractionated schedule changes what a quoted tolerance actually means.
Staging drives target volumes
T and N stage set both the gross and the elective volume. In head and neck, gross disease often receives 70 Gy while elective nodal levels receive 50–54 Gy in 2 Gy fractions; in prostate, T stage, PSA and Gleason score decide whether pelvic nodes are covered. Concurrent chemotherapy tightens mucosal and pulmonary constraints rather than loosening them, and palliative intent changes the arithmetic entirely (8 Gy × 1, 20 Gy in 5, 30 Gy in 10).
Tolerances are fractionation-dependent
QUANTEC-style numbers assume 2 Gy fractions: spinal cord Dmax 45–50 Gy, brainstem 54 Gy, optic chiasm 55 Gy, parotid mean 26 Gy, lung V20 ≤30–35% with mean lung dose ≤20 Gy, heart mean ≤26 Gy in breast planning, liver mean ≤30 Gy, bilateral kidney mean ≤18 Gy. Convert before comparing schedules: BED = nd(1 + d/(α/β)), EQD2 = BED/(1 + 2/(α/β)). Typical α/β is about 1.5 for prostate, 3–4 for breast and late rectal injury, and about 10 for mucosa and most tumour responses.
Worked example - is a hypofractionated prostate course hot or cold?
Compare 78 Gy in 39 fractions with 60 Gy in 20 fractions of 3 Gy. For tumour (α/β = 1.5): 78 × (1 + 2/1.5) = 182 Gy₁.₅, versus 60 × (1 + 3/1.5) = 180 Gy₁.₅ - effectively isoeffective. For late rectum (α/β = 3): 78 × (1 + 2/3) = 130 Gy₃ versus 60 × (1 + 3/3) = 120 Gy₃, so the 20-fraction course is biologically gentler to late tissue. The marks come from stating the α/β used and noting the comparison only holds while the same α/β is assumed to apply.
Traps that cost marks
Quoting a constraint without its metric (Dmax versus D0.03 cc versus mean versus V20). Lifting a 2 Gy tolerance into SBRT without EQD2 conversion. Approving CTV coverage while the PTV is underdosed, or judging a plan from one axial slice instead of the DVH and full 3D distribution. Forgetting that cord tolerance is a cumulative, life-long number that includes any prior course.
Radiation Physics, Beam Data, and Treatment Delivery
Domain 2 tests the physics you apply daily: how dose builds and attenuates, how geometry and energy change output, and where photon and electron behaviours diverge at the patient surface.
Photon dose behaviour
Percent depth dose climbs to dmax (roughly 1.5 cm at 6 MV, 2.5 cm at 10 MV, 3.5 cm at 18 MV) then falls; build-up explains skin sparing and why bolus is used for chest wall and scar. PDD depends on SSD and field size, while TMR is essentially SSD-independent and is the correct parameter for isocentric (SAD) calculations. When SSD is extended for a large patient, correct the PDD with the Mayneord factor: F = [(SSD₂ + dmax)/(SSD₁ + dmax)]² × [(SSD₁ + d)/(SSD₂ + d)]²
Electrons
The practical range is about E/2 in centimetres and the 90% therapeutic depth about E/4, so 6 MeV treats to roughly 1.5 cm beyond which dose falls steeply. Output depends on cone size and SSD (virtual source distance is around 95 cm), so inverse-square corrections are mandatory when the standoff changes.
Worked example - extended SSD correction
The PDD at 10 cm for a 10 × 10 cm 6 MV beam is 67.0% at 100 cm SSD. At 120 cm SSD: F = (121.5/101.5)² × (110/130)² = 1.4329 × 0.7160 = 1.026. The corrected PDD is 67.0 × 1.026 = 68.7%, and the monitor units fall accordingly. Note that the projected field size at the patient also grows with distance, so a collimator setting that gave 10 × 10 at 100 cm gives about 12 × 12 at 120 cm unless the jaw setting is reduced.
Traps that cost marks
Using PDD at an extended SSD without the Mayneard correction above. Confusing dmax with the depth of maximum surface dose on an unbolused tangent field. Treating electron output as constant across cones, or ignoring the virtual source distance. Mixing up collimator field size with the field size projected at depth.
External Beam Treatment Planning and Dose Calculation
Domain 3 is the core of the exam: volume definition, margin design, dose algorithms, and objective plan evaluation. It tests whether you understand what the planning system is doing rather than which buttons to press.
Volume definitions follow ICRU 83
GTV is visible or clinically palpable disease; CTV adds suspected microscopic spread; ITV encompasses CTV motion across the respiratory cycle; PTV adds setup and motion margin; PRV is an expanded organ-at-risk used for planning. The van Herk margin recipe separates uncertainty types: M = 2.5Σ + 0.7σ, where Σ is the total systematic error and σ the random error.
Algorithms and heterogeneity
The choice of algorithm changes numbers in lung and near air cavities. Pencil-beam convolution can under-dose low-density lung relative to superposition, Acuros and Monte Carlo. Acuros reports dose-to-medium while AAA reports dose-to-water, so cross-comparisons need that context, not a naked percentage difference.
Worked example - margin from error data
A clinic measures systematic error Σ = 3 mm (CT-to-cone-beam registration, couch, laser) and random error σ = 2 mm (intrafraction motion, daily setup). M = 2.5(3) + 0.7(2) = 7.5 + 1.4 = 8.9 mm, so a 9 mm CTV-to-PTV margin is justified on data, not habit. If a compression belt or daily CBCT reduces σ to 1 mm, the margin falls to 8.2 mm; if Σ rose to 5 mm because lasers drifted, tolerance rose to 13.9 mm and the physics issue must be fixed rather than absorbed into the margin.
Evaluating the plan
Report coverage (D95, D98, V100), homogeneity (D2 − D98 divided by D50), conformity, gradient index and each OAR metric stated in the prescription. A PTV V95 of 100% with a hot D2 of 115% is a worse plan than 97% coverage with a D2 of 107% in most disease sites.
Traps that cost marks
Assuming prescription dose means 100% PTV coverage is required. Comparing algorithm doses without noting dose-to-medium versus dose-to-water. Ignoring field-junction doses in craniospinal or head-and-neck matching. Forgetting that PRV expansions for cord and brainstem must be respected in the optimization, not just the report.
Brachytherapy Physics and Planning
Domain 4 tests source dosimetry, the TG-43 formalism, and the clinical metrics that decide whether an implant is acceptable. Expect questions that mix source decay arithmetic with anatomy-specific reporting parameters.
TG-43 formalism
Dose around a sealed source is calculated as D(r,θ) = Sk · Λ · [G(r,θ)/G(r₀,θ₀)] · g(r) · F(r,θ), where Sk is air kerma strength (1 U = 1 cGy·cm²·h⁻¹ = 1 µGy·m²·h⁻¹), Λ the dose-rate constant in cGy·h⁻¹·U⁻¹, G the geometry factor, g the radial dose function and F the anisotropy function. TG-43 assumes a homogeneous water medium and does not model applicator or tissue heterogeneity; model-based algorithms (TG-186) exist for that reason.
Sources at a glance
Ir-192 HDR: half-life 73.8 days, average energy roughly 0.38 MeV, treated with afterloaders in fractions. I-125: 59.4 days, ~27 keV, prostate monotherapy 144 Gy. Pd-103: 17 days, ~21 keV, 125 Gy. Cs-131 has a shorter 9.7-day half-life and is used in some permanent implants.
Clinical reporting metrics
Prostate LDR: post-implant D90 and V100, with V100 above about 90% and D90 near or above prescription. Cervix HDR (GEC-ESTRO/EMBRACE style): high-risk CTV D90 ≥ 85 Gy EQD2, with D2cc limits around 90 Gy EQD2 bladder, 75 Gy rectum and sigmoid, 80 Gy bowel.
Worked example - decay-adjusted dwell time
An Ir-192 prostate HDR plan prescribes 15 Gy with Sk = 40,000 U and a total dwell time of 300 s. Treatment happens after decay to 38,000 U. Dose rate scales with Sk, so dwell time must scale by 40,000/38,000 = 1.053, giving 316 s. Confirm against the treatment planning system and the independent second check - never hand-scale alone. Note the source loses half its strength every 73.8 days, so a week of delay costs about 6%.
Traps that cost marks
Confusing apparent activity (mCi) with air kerma strength (U). Omitting the anisotropy function for low-energy seeds. Forgetting that dwell position count and step size change the dose distribution even at fixed total dwell time. Applying TG-43 near a shielded applicator without acknowledging the approximation.
Quality Assurance and Radiation Safety
Domain 5 combines machine and plan QA tolerances with radiation protection rules. The examiner wants the correct tolerance, the correct frequency and the correct action level - not a vague statement that QA was performed.
TG-142 style tolerances
Daily: output constancy within 3%, lasers within 2 mm, optical distance indicator within 2 mm, imaging and collision interlocks functional. Monthly: output calibration within 2%, PDD/JDD constancy within 1%, MLC leaf position within 1 mm. Annual: mechanical and radiation isocenter within 1 mm, comprehensive output and energy checks, imaging dose review. MLC and gantry checks should use the same phantom and be trended, because drift matters more than a single pass.
Patient-specific and plan QA
IMRT/VMAT verification commonly uses gamma analysis at 3%/3 mm with a 95% pass rate above a 10% dose threshold, with tighter 2%/2 mm criteria where gradients permit. Report global versus local normalization explicitly, because the same data can pass one and fail the other. Independent MU or dose second checks are separate from gamma results and should be documented for every plan.
Dose limits (occupational and public)
Effective dose for radiation workers 50 mSv per year (20 mSv averaged), lens of eye 150 mSv per year, extremities and skin 500 mSv per year, declared pregnant worker 0.5 mSv in any month, and members of the public 1 mSv per year. ALARA applies on top of the limits: badge readings, area surveys, and time-distance-shielding controls.
Worked example - action level reasoning
A monthly output check reads 1.028 cGy/MU against a baseline of 1.000, a 2.8% deviation. That is inside the daily 3% constancy band, but exceeds the 2% monthly calibration tolerance, so the linac is scheduled for calibration and clinical output is re-verified before the next treatment day. Trend data showing two prior months at +1.8% and +2.3% strengthens the case that this is drift, not measurement noise.
Traps that cost marks
Mixing daily and monthly tolerances (2% is monthly calibration, 3% is daily constancy). Reporting gamma results without stating normalization or threshold. Relying on a single phantom measurement when the ion chamber calibration certificate has expired.
Informatics, Data Flow, and Professional Practice
Domain 6 covers the data that surrounds the plan and the professional obligations around it: DICOM objects, image registration, transfer integrity, and the dosimetrist's scope of practice.
DICOM RT objects
RT Structure Set holds contours and ROI names; RT Plan holds beams, MLC, prescription and fractionation; RT Dose holds the dose grid and DVH data; RT Image holds reference images. Coordinate data are patient-based in the DICOM LPS frame, and every object carries a FrameOfReferenceUID that must match across the CT, structure set and plan. Modern workflows also carry RT Radiation Set and RT Radiation instances, but the classic trio remains the safest mental model.
Registration and fusion
Rigid registration is appropriate for brain and prostate; deformable registration (DIR) is used for re-planning and adaptive work but must be visually validated structure by structure, especially at the skull base, bowel and bladder. Fusion accuracy directly sets GTV delineation accuracy, so a 3 mm registration error can dominate every downstream margin.
Worked example - the 3 mm isocenter shift
A boost plan is exported to the record-and-verify system and comes back with the isocenter 3 mm posterior. The CT had been re-imported after a re-scan, creating a second FrameOfReferenceUID, and the plan referenced the first while the structures referenced the second. The fix is not to nudge the isocenter by hand: compare FrameOfReferenceUID and ImagePositionPatient tags across all objects, re-export from the treatment planning system, and confirm with a phantom end-to-end test before treating.
Scope, consent and error handling
A CMD works under the supervision of a radiation oncologist and does not prescribe, alter prescription or approve a plan on their own authority. Consent, treatment discussion and medical decisions belong to the physician. Errors - wrong laterality, wrong fractionation, dose delivered to a mislabelled structure - are reported through the departmental incident system and may be entered into a national learning system; concealment is a professional failure, not a QA step. Patient information stays within HIPAA-compliant systems, and images used for teaching must be de-identified.
Traps that cost marks
Assuming a plan opens correctly because the patient name matches. Ignoring duplicated ROI names when a structure set is copied. Treating DIR output as ground truth. Confusing the dosimetrist's plan-optimization responsibility with physician prescribing authority.