Impact of dopants on X‑ray attenuation of 3D‑printed polymers for mammography and tomosynthesis spectra

Abstract Background 3D‐printing enables the production of anthropomorphic breast phantoms for assessing image quality in digital mammography (DM) and breast tomosynthesis (DBT). 3D‐printing materials must be selected carefully to mimic breast tissue attenuation for clinical spectra. Purpose This study investigates the impact on the effective X‐ray attenuation coefficient ( µ eff ) of polymer formulation and dopants (e.g., colorants) used in 3D‐printing materials. It also analyses the impact on their equivalent glandular ratios ( BR ) relative to commercial materials (CIRS). Methods Five base polymers (PLA, ABS, PETG, HIPS, RESIN) were selected among the catalogue of five manufacturers. A collection of nineteen variants (base polymer + color) was 3D‐printed as step‐wedges with thicknesses ( t ) from 0.5 to 5.5 cm and imaged on a clinical mammography system (DM: W/Rh with the anti‐scatter grid, DBT: W/Al without grid, at 27, 29 and 31 kVp). The µ eff (t) values were derived from the mean pixel value measured at each step of the logarithmically transformed images. As reference, µ eff values were measured for the CIRS materials with well‐known BR . A two‐parameter empirical model was used to fit the values as: µ eff (t) = µ 0 /(1+kt) , where µ 0 corresponds to µ eff at thickness tending to zero, and k is the decrease rate with thickness due to beam hardening and scatter. The impact of dopants on polymer attenuation was evaluated by analyzing both µ eff and µ 0 values. Based on µ 0 –BR relationships of CIRS materials, the corresponding BR values were computed for the investigated materials. Results For all acquisition conditions, some color variants of 3D‐printed materials with the same base polymer exhibit noticeable differences in µ eff , independently of the thickness, particularly for DM. For this modality, the maximum differences within µ eff values of materials sharing polymer base decreased by 62% for PLA‐based, 44% for PETG‐based and 30% for ABS‐based materials, when the thickness increases from 1.5 to 5.5 cm. For DBT, a lower variability of µ eff values was observed. The materials best mimicking purely glandular and adipose tissues were selected based on µ eff , µ 0 and BR results. Specifically, relative differences in µ 0 (averaged across kVp), reported as paired values (DM, DBT) were: (−3%, +1%) and (−1%, +4%) for PLA crystal silver and PLA blue versus CIRS with BR = 100% (glandular tissue); (+3%, +1%) for ABS white versus CIRS with BR = 0% (adipose tissue). According to the BR values derived for all acquisition conditions, PLA blue and PLA crystal silver mimicked glandular tissue ( BR : 101%–109%, 95%–100%, respectively), while ABS white mimicked adipose tissue ( BR : −2% to 8%). PETG translucent blue and RESIN corresponded to intermediate glandular ratios ( BR ≈ 50%–70%). Conclusion 3D‐printed materials sharing the same polymer base can exhibit substantially different attenuation for DM/DBT spectra, resulting in different BR equivalences due to formulation variations (e.g., pigments and dopants). Therefore, phantom developers should not rely solely on base polymer designations and declared density when selecting materials for production of breast phantoms; instead, the attenuation of each specific formulation needs to be verified and periodically re‐evaluated under the intended mammography conditions.