Quality assurance of 3D-printed patient specific anatomical models: a systematic review.

Abstract

Background: The responsible use of 3D-printing in medicine includes a context-based quality assurance. Considerable literature has been published in this field, yet the quality of assessment varies widely. The limited discriminatory power of some assessment methods challenges the comparison of results. The total error for patient specific anatomical models comprises relevant partial errors of the production process: segmentation error (SegE), digital editing error (DEE), printing error (PrE). The present review provides an overview to improve the general understanding of the process specific errors, quantitative analysis, and standardized terminology.

Methods: This review focuses on literature on quality assurance of patient-specific anatomical models in terms of geometric accuracy published before December 4th, 2022 (n = 139). In an attempt to organize the literature, the publications are assigned to comparable categories and the absolute values of the maximum mean deviation (AMMD) per publication are determined therein.

Results: The three major examined types of original structures are teeth or jaw (n = 52), skull bones without jaw (n = 17) and heart with coronary arteries (n = 16). VPP (vat photopolymerization) is the most frequently employed basic 3D-printing technology (n = 112 experiments). The median values of AMMD (AMMD: The metric AMMD is defined as the largest linear deviation, based on an average value from at least two individual measurements.) are 0.8 mm for the SegE, 0.26 mm for the PrE and 0.825 mm for the total error. No average values are found for the DEE.

Conclusion: The total error is not significantly higher than the partial errors which may compensate each other. Consequently SegE, DEE and PrE should be analyzed individually to describe the result quality as their sum according to rules of error propagation. Current methods for quality assurance of the segmentation are often either realistic and accurate or resource efficient. Future research should focus on implementing models for cost effective evaluations with high accuracy and realism. Our system of categorization may be enhancing the understanding of the overall process and a valuable contribution to the structural design and reporting of future experiments. It can be used to educate specialists for risk assessment and process validation within the additive manufacturing industry.