High-fidelity Monte Carlo dosimetry of radiosensitive organs in head and neck CT: A virtual TLD study with comprehensive statistical uncertainty analysis.

PURPOSE: To quantify absorbed doses to six critical head and neck organs (occipital lobe, temporal lobe, hypothalamus, thalamus, parotid gland, and both eyes) during head CT using high-fidelity Monte Carlo simulation as an ethical alternative to in vivo dosimetry.

METHODS

A voxelized phantom was generated from DICOM images of the RANDO phantom. A realistic X-ray tube (135 kVp) was modeled in GEANT4 (v11.1.3) to simulate clinical head CT protocols. Seventy-four virtual GR-200 thermoluminescent dosimeters (TLDs) were placed at predefined organ coordinates, and 2 × 108 primary photons were tracked. Descriptive statistics (mean, SD, SEM, 95% CI, CV%) and non-parametric tests (Kruskal-Wallis followed by Mann-Whitney U with Bonferroni correction) were applied to the raw dose distributions.

RESULTS

The parotid gland received the highest mean dose (25.83 mGy) and the right eye the lowest (19.10 mGy). The temporal lobe exhibited the smallest variability (CV = 0.39%), while the right eye showed the largest (CV = 11.52%). The Kruskal-Wallis test revealed highly significant differences among organs (H = 62.34, p = 1.5 × 10-11). Pairwise comparisons indicated that the parotid gland dose was significantly higher than both eyes (p ≈ 3.2 × 10-4), whereas no significant difference existed between the two eyes (p = 0.86) or between occipital and temporal lobes (p = 0.43). Dose ratios relative to the thalamus ranged from 0.824 (right eye) to 1.114 (parotid gland).

CONCLUSION

The statistically robust organ-dose estimates validate the GEANT4-based model and align with previous experimental data. Monte Carlo simulation with strategically placed virtual TLDs offers a powerful, ethical tool for precise dose assessment, providing a valuable reference for protocol optimization and radiation protection in head and neck CT.