Purpose or Case Report: 3D printed models from MRI scans can effectively demonstrate the surface structure of the brain. Previous workflows focus on adult brains as a basis for prints. Our database of pediatric MRI brains who had perinatal hypoxic ischemic injury and presented late for imaging, with pathology causing cortical surface irregularities and parenchymal cysts. Difficulties arise in accurate depiction of the cortex on 3D print models in these pathologic brains. We aim to demonstrate effective workflows to accurately and efficiently print 3D models of especially pathologic pediatric MRI brains. Also, to critically and empirically test and refine the various steps involved in producing 3D print models which include segmentation of the MRI volume into tissue classes, generation of a surface model from this volume, preparation and final print of a 3D model.
Methods & Materials: 35 MRI brain scans of patients presenting with hypoxic ischemic injury were used and preliminary reads enabled categorization of the scans into normal, atrophy, cavity and other. Initial segmentation performed then allowed further refinement of variable settings and evaluation of other methods of segmentation, to generate segmentation models. Mesh surface generation was then performed with further refinements, using various application extensions of Matlab, Mathworks. Print preparation involved surface mesh repairs, removal of internal structures and surface smoothing. Comparison was also made between printers/ printing technologies.
Results: Initial segmentation produced only 20 out 35 (57.1%) usable models. Further refinement increased this to 31 out of 35 (91.4%). The best quality segmentation models for normal and other patients were produced by CAT12 segmentation. The best quality depictions were created using SPM12.

FDM printing was found to be more accessible and economic although all print methods produced high quality models. Print settings were found to be optimal at 0.2mm layer height and 5% infill.
Conclusions: We showed improved MRI segmentation success rate for pathological pediatric brains as compared to using default, previously successful for segmenting adult brains. Through methodically testing various combinations of parameters, ideal settings have been determined for improving segmentation success rate and quality. High quality models were printed with compromise between reducing printing time and maintaining quality and structural strength of the models.