J. Day1, A. Espinoza Orias3, M. Supanich2, M. Kocak2, K. Stenson4, V. Stubbs4, M. Akyuz2, S. Wrenn5 1Rush University Medical Center, Chicago, IL, USA 2Rush University Medical Center, Radiology And Nuclear Medicine, Chicago, IL, USA 3Rush University Medical Center, Orthopedic Surgery And Radiology, Chicago, IL, USA 4Rush University Medical Center, Otolaryngology And Head And Neck Surgery, Chicago, IL, USA 5Rush University Medical Center, Endocrine Surgery, Chicago, IL, USA
Introduction: Ultrasound (US) of the neck is the standard of care for patients with goiter (thyroid enlargement) to measure gland dimensions, determine disease severity and develop a treatment plan. US limitations include operator variability, spatial heterogeneity and irregularity, transducer size being smaller than the goiter/nodule, and substernal projection. These can impair goiter visualization and subsequent volume calculation. The purpose of this proof-of-concept study is to investigate the efficacy of creating 3D-printed thyroids based on computed tomography (CT) imaging as a means of accurate thyroid volume calculation and visualization.
Methods: 10 retrospective cases were used in this study: 5 goiter cases and 5 control non-pathological thyroid cases. CT neck imaging with reconstructed slice thickness of 1mm or thinner was performed from June-July 2024 and 9 of the 10 patients received intravenous contrast. For each case, a neuroradiologist used CT neck imaging to calculate thyroid volume utilizing the volumetric ellipsoid method (length x width x height x a correction factor of 0.524 for each lobe). The resulting DICOM slices were segmented with commercially available software (Mimics 26.0), yielding virtual 3D thyroid models and their corresponding calculated volumes. Finally, volumes of printed models were measured using the Archimedes method (water displacement).
Results: Paired t-tests demonstrated there was not a significant difference between radiologist-predicted and Mimics-predicted goiter volumes, nor between radiologist-predicted and Mimics-predicted control volumes (significance defined as p<0.05). The same tests showed there was not a significant difference between Mimics-predicted goiter volumes and 3D goiter volumes measured by the Archimedes method, nor between Mimics-predicted control thyroid volumes and 3D control thyroid volumes measured by the Archimedes method (significance defined as p<0.05). The mean print time was 6.17 hours per model (6.55 for goiters, 5.8 for normal thyroids). The mean cost of material used was $5.95 per model ($9.44 for goiters, $2.47 for normal thyroids).
Conclusion: While this pilot study cannot demonstrate technique superiority, the results suggest 3D-printed thyroids provide a similarly accurate estimate of volume compared with traditional radiology estimates. The study may be underpowered to detect volume differences between techniques and additional investigation is required. Future directions for this research group include a prospective, randomized trial that compares 3D printed volume to both US measures and those of excised thyroid tissue. We plan to qualitatively explore additional benefits of 3D printing such as intervention planning and patient education.