Dosimetry and Treatment Calculation

Radiation dose is the amount of energy deposited per unit mass. Gray (Gy) is SI unit: 1 Gy = 1 joule of energy per kilogram of matter. In oncology, doses are smaller so centiGray (cGy) is often used: 1 cGy = 0.01 Gy = 1 rad (old unit, still used colloquially). Example: breast cancer prescription = 5000 cGy = 50 Gy over 25 fractions = 200 cGy per fraction. Prostate low-risk cancer = 6400 cGy = 64 Gy over 32 fractions = 200 cGy per fraction. Stereotactic radiosurgery to brain lesion = 1500 cGy = 15 Gy in 1 fraction. Effective dose (sievert, Sv) is different: it accounts for biological effect of radiation type (photon = 1 Gy ~ 1 Sv; proton = different weighting); not used in treatment planning. Confusion between Gray (dose) and sievert (effective dose) can cause errors; always verify units. Dose reporting in medical records must be clear: cGy vs Gy, total dose vs per-fraction dose.

Inhomogeneity = variation in tissue density (lung is ~0.3 g/cm³ vs muscle ~1.0 g/cm³). When a photon beam passes through lung, it attenuates less (fewer interactions) so the dose is lower than in muscle for the same fluence. Early dose algorithms (water-only convolution) ignore inhomogeneity and overestimate dose distal to lung (behind lung). Modern algorithms (convolution, superposition, Monte Carlo) apply density corrections. Example: 10 MV photon beam at depth d in water = X dose; same depth in lung = 0.7×X (lower); distal to lung (same depth in muscle) = higher than water because the reduced attenuation in lung allows more dose to penetrate. Accuracy matters: 5-10% dose errors are common at lung interfaces without correction. Heterogeneity is especially important in: lung cancers (tumor in low-density lung), mediastinal tumors (heterogeneous: lung, mediastinum, heart), and head/neck (bone, soft tissue, air sinuses). Algorithms vary in accuracy; Monte Carlo is gold-standard but slow; convolution/superposition is fast and accurate enough for clinical use. Check your TPS algorithm and understand its limitations.

Output measurement verifies the linac delivers the correct dose rate. Farmer chamber (cylindrical ion chamber, 0.125 or 0.6 cc volume) is placed in water phantom at reference depth (typically 10 cm for 10 MV, 5 cm for electron), at 100 cm source-to-surface distance (SSD), in a standard field (10×10 cm²). Linac delivers a monitor unit setting (typically 100 MU). Charge is collected by electrometer and converted to dose using calibration factor. Protocol (AAPM TG51): (1) condition chamber (pre-irradiate), (2) set up geometry precisely (SSD, depth, field size), (3) irradiate, (4) read charge, (5) apply temperature/pressure correction, (6) apply ion recombination correction, (7) calculate dose. Expected output: 1 cGy per monitor unit (or specified institutional standard). If measured ≠ expected by >1%, investigate (wrong field size, chamber positioned incorrectly, chamber damaged, linac calibration off). Output must be checked daily or before patient treatment to ensure consistency. Drift >2% triggers service call. Understanding the protocol is essential for QA.

Point dose: measure dose at a single point (e.g., ion chamber at isocenter) and compare to calculated dose. Fast, sensitive to major errors (wrong field size, wrong energy), but misses off-axis errors. Setup: chamber at known position, calculate expected dose for that geometry/fluence, irradiate, read charge, convert to dose, compare. Pass/fail criterion: ±3-5% depending on institutional tolerance. Advantage: simple, quick. Disadvantage: one point may miss dose errors elsewhere. 2D dose: expose radiochromic film (EBT3, EBT4) or electronic detector array (portal dosimetry) to dose distribution, digitize image, compare 2D calculated vs measured dose. Analyzed using gamma index: voxel passes if it is within 3% dose AND 3 mm distance from nearest calculated point. Example: 95% of measured voxels pass gamma → plan accepted. Advantage: spatially resolved, catches dose errors across field. Disadvantage: analysis is complex, requires calibration, film must be scanned. 3D dose: place dosimeter array (ion chamber array or alanine pellets in 3D arrangement) in phantom, irradiate, read doses, reconstruct 3D distribution, compare to calculated. Most comprehensive; detects errors in all three dimensions. Disadvantage: time-consuming, expensive, not routine. Standard: point dose + 2D 2D dose verification is minimally required before patient treatment.

Dose-volume histogram (DVH) is a graph showing what volume of a structure receives what dose. X-axis = dose (Gy), Y-axis = volume % (0-100%) or absolute volume (cc). Cumulative DVH: shows percent of volume receiving at least that dose. Differential DVH: shows incremental volume in dose bins. Example: prostate DVH showing 100% of PTV receives >64 Gy (coverage), 50% of bladder receives >50 Gy, 30% of rectum receives >50 Gy. Plan evaluation uses DVH to check: (1) target coverage (PTV DVH: 95-107% of prescribed dose in ≥95% of volume), (2) organ constraints (e.g., rectum: <50% receives >50 Gy, <25% receives >65 Gy, <1% receives >75 Gy), (3) dose homogeneity (is dose uniform within target or is there cold spots?). DVH simplifies complex 3D data into a 1D plot; downside is loss of spatial information (DVH cannot distinguish if 50% rectum is in cranial or caudal region, which matters for late toxicity). DVH-based plan optimization is now standard; many TPS use DVH constraints to auto-optimize beams. DVH is a tool; clinical judgment and anatomy review must complement it.

Reference dose is the dose at a reference point (usually isocenter or a point inside the target) used as the prescription basis. Prescription example: 'Deliver 70 Gy to the planning target volume (PTV) at 2 Gy per fraction' means the reference dose is 70 Gy. ICRU (International Commission on Radiation Units) defines reference point as a point inside the target where dose is near the average of the target. For IMRT, reference dose is often the mean dose to PTV (different from dose at a point). Why it matters: dose reporting clarity. If report says '70 Gy delivered' but doesn't specify reference point, clinician may misinterpret (was it the maximum dose? minimum? mean?). ICRU 83 standardizes reporting: specify prescribed dose, reference dose, dose at ICRU point (if clinically relevant), and DVH summary. Ambiguous dose reporting can lead to overdose (patient got more than intended) or underdose (patient got less, tumor underdosed). Dose escalation (giving more dose to improve outcomes) requires clarity: is the escalation compared to the previous reference dose or a new reference? Documentation is critical to avoid errors.

Monte Carlo (MC) is a stochastic algorithm that simulates the random interactions of individual photons and electrons with tissue. Each photon is tracked: where it originates, what direction, how it interacts (pair production, Compton scattering, photoelectric effect), where secondary electrons go, and how energy is deposited. Averaging millions of photon histories gives dose distribution. Advantages: handles complex geometry (bone, lung, metal), very accurate, can model irregular fields. Disadvantage: slow (minutes to hours per plan vs seconds for pencil beam). When necessary: (1) high-Z materials (titanium implants, dental fillings) cause artifacts in convolution algorithms; MC is more accurate, (2) low-density regions (air cavities in head/neck, lung) where heterogeneity effect is large, (3) small fields (<1 cm) where lateral scatter is important, (4) electron beams in heterogeneous media, (5) brachytherapy near shielded sources. Many modern TPS use fast MC (GPU-accelerated) to reduce computation time. For routine cases in homogeneous media (simple prostate, uniform breast), convolution is adequate and faster. Physicist must understand algorithm choice trade-offs: speed vs accuracy.

Absolute dose calibration: linac output is set so 1 monitor unit = 1 cGy at reference conditions (typically SSD 100 cm, depth 10 cm, 10×10 field, 10 MV photons). Uses reference standard: 60Co beam or reference linac calibrated against traceable standard (national standard). Farmer chamber protocol TG51 defines the procedure. Absolute calibration is essential: if linac is under-calibrated by 5%, all patient doses are under by 5% (tumors underdosed, normal tissue spared but underdose worsens outcome). If over-calibrated by 5%, all doses are over (tumors better covered but normal tissue toxicity increases). QA checks calibration monthly and after servicing. Relative dose: comparison of dose at point A to dose at point B (both in same patient or phantom). Example: dose at depth 10 cm vs depth 15 cm (depth dose curve). Relative dose is independent of calibration; it reflects beam quality and geometry. Dose distributions (isodose lines, DVH) are based on relative dose. Clinical errors: measuring absolute dose without proper calibration = wrong dose prescription; relative comparisons (e.g., 'is this new linac similar to old linac?') can be done with relative dose measurements. Understanding absolute vs relative is critical for QA and error detection.