Preclinical feasibility of EPID-based transit dosimetry for total body irradiation

Conventional extended-distance total body irradiation (TBI) is commonly verified using point dosimeters, which provide limited spatial information and require additional handling during a prolonged treatment procedure. A portable charge-coupled device (CCD)-based electronic portal imaging device (EPID) positioned downstream of the patient may provide wider-area transit information in a geometry in which a gantry-mounted EPID cannot readily be used.

To develop and perform preclinical phantom validation of a CCD-EPID method for transit dose estimation during fixed-field long-SSD TBI and to identify the calibration and normalization factors relevant to subsequent patient validation.

A portable CCD-based EPID was positioned at 400 cm, with water-equivalent or anthropomorphic phantoms centered at 350 cm. Direct-irradiation characteristics were evaluated for 4- and 10-MV photon beams, including signal response to dose-rate/delivered MU, field size, short-term frame reproducibility, and in-plane response. For ETD calibration, 6-36-cm-thick water-equivalent phantoms were irradiated with 10-MV photons using 1000 MU at 600 MU/min and a 10 × 10 cm² jaw setting at isocenter, corresponding to approximately 35 × 35 cm² at the phantom and 40 × 40 cm² at the EPID. Summed EPID signals were related to midplane absorbed dose measured using a Farmer-type ionization chamber. ETD was independently compared with radiophotoluminescent dosimeters (RPLDs) in 12-, 24-, and 36-cm slab phantoms and at 11 locations in a RANDO phantom. Four separate RANDO irradiations were centered on the head, chest, abdomen, and pelvis.

The 10-MV EPID signal was highly linear with dose rate/delivered MU (r = .999) and field size (r = .995). Flatness was 1.4% and 1.3%, and symmetry was .3% and 1.4%, in the ceiling-floor and gun-target directions, respectively. The summed EPID signal-to-ionization-chamber dose relationship showed R² = .999. ETD differed from RPLD dose by -1.6%, -1.4%, and .0% in the 12-, 24-, and 36-cm slab phantoms, respectively, within the expanded RPLD uncertainty. In the RANDO phantom, ETD was consistently higher than RPLD dose, with a mean relative difference of 7.7 ± 3.9%. Because calibration and RANDO verification used the same field size and the detector showed high dose/MU linearity, a field-size mismatch or gross signal nonlinearity is unlikely to be the dominant explanation. Heterogeneous attenuation and scatter, local path-length/ROI correspondence, and session-dependent CCD sensitivity are plausible contributors.

This preclinical phantom study demonstrates the feasibility of wide-area CCD-EPID transit-dose estimation for conventional long-SSD TBI. Agreement in homogeneous phantoms supports the calibration approach, while the positive RANDO bias identifies the need for same-day sensitivity normalization and further validation in repeated anthropomorphic, heterogeneous, lung-shielded, and full-clinical-field conditions before patient application.