[The validation of stereolithographic anatomical replicas: the authors' own experience and a review of the literature].

The visual three-dimensional (3D) reconstruction of CT findings has been used since the Seventies to design and plan complex surgical procedures. The availability of such models and the development of computer science have permitted, since the mid-Eighties, the medical use of rapid prototyping for anatomical modelling. We studied the technical steps of CT data processing for rapid prototyping and the dimensional and structural accuracy of replicas of skeletal components relative to the originals. A dried mandible and an arthrotic hip joint were compared with their stereolithographic replicas using the measurements made on CT images. The 3D graphic models were processed with a commercially available software and replicated with an SLA 250 stereolithographer (3D System Inc, Valencia CA, USA). Satisfactory morphologic agreement was found between the original and its replica. The mandibular replica exhibited dimensional errors ranging 0 to 4.03%, z-axis shortening and an increase in gonial angle, but with a reliable reproduction of details. The hip replica exhibited dimensional errors ranging .2% to 8.53%; it appeared to be less affected than the mandibular replica by geometrical distortion but it was less accurate in the reproduction of cancellous bone and arthrosis. The accuracy of the solid replicas depends, in the different manufacturing steps, on several factors, namely: during data acquisition, the accuracy and suitability of original data, depending in turn on equipment and examination parameters--especially the algorithm, acquisition time, gantry tilt and slice thickness; image thresholding and manipulation procedures during graphic model creation; data transfer, creation of the necessary supports and replica finishing during manufacturing; residual polymerization affected by environmental agents and preservation conditions during maintenance.