CT-based radiodensity validation of a custom orbital phantom: quantitative comparison with the CIRS ATOM reference and clinical data.

In this study the development and radiological validation of a custom-designed orbital anthropomorphic phantom was investigated. The phantom was intended to accurately reproduce the radiodensity of key oculo-orbital tissues in computed tomography (CT). The phantom was fabricated using epoxy-resin-based composites modified with calcium carbonate, acetone, and sodium bicarbonate to achieve tissue-equivalent attenuation properties. Four representative tissues-orbital bone, crystalline lens, extraocular muscle, and optic nerve-were simulated to obtain optimized material compositions. CT images were produced on a clinical scanner under standardized conditions (80-120 kV), and Hounsfield Unit (HU) values were measured using a consistent region-of-interest (ROI) methodology. The results were compared with those obtained from a commercial reference phantom (CIRS ATOM Adult Phantom) and a clinical dataset of 256 adult brain CT images, providing statistically robust reference values. The developed phantom demonstrated strong agreement with clinical data, particularly for soft tissues, with relative differences of - 3.83% for the eye lens, - 3.63% for muscle, and + 3.49% for the optic nerve. In contrast, the commercial phantom showed significant discrepancies for soft tissues, including - 48.31% for the eye lens and + 30.62% for the optic nerve, due to its homogeneous material composition. Both phantoms accurately reproduced bone radiodensity, with relative differences remaining below 5%. Furthermore, the developed phantom preserved tissue-specific HU differentiation across varying tube voltages, reflecting realistic energy-dependent behavior consistent with X-ray interaction mechanisms. Overall, the developed phantom provides a realistic radiological representation of the orbital region and a promising experimental platform for CT imaging and dosimetric applications, while further validation across different CT systems and acquisition protocols is warranted to establish its broader applicability.