3D-Printed Patient-Specific Spinal Cord Injury Models: Design, Implementation, Evaluation

Thesis Abstract

3D-Printed patient-specific spinal cord injury models hold promise for advancing understanding of injury mechanics, surgical planning, and rehabilitation by providing anatomically accurate, end-to-end simulacra of individual cervical and thoracic lesions. The study addresses the gap between generic cadaveric or synthetic models and patient-specific variability that hampers translational insights in spinal cord injury (SCI) research and clinical training. The aim is to design, implement, and evaluate a workflow for producing high-fidelity, patient-specific 3D-printed SCI models that integrate biomechanical properties, neural tissue proxies, and removable lesion components to support experimental assessment, surgical rehearsal, and education. Specific objectives are to (1) develop a robust imaging-to-model pipeline using high-resolution MRI/CT data to generate individualized spinal anatomy and lesion geometries, (2) fabricate multimaterial 3D-printed models with calibrated mechanical properties mimicking vertebral, intervertebral, and spinal cord tissues, (3) embed modular lesion features (compressive, contusive, and transection patterns) and assess their replicability across models, (4) validate geometric and mechanical fidelity against cadaveric benchmarks via surface scanning and dynamic indentation testing, and (5) evaluate educational and planning utility through a mixed-methods study involving 20 spine surgery trainees and 10 seasoned surgeons using task-based performance metrics and structured interviews guided by the Theory of Situated Learning and the Cognitive Load Theory. A cross-disciplinary mixed-methods design is adopted. The population comprises adult human cadaveric spine segments (C5–T1) and clinical imaging datasets from 40 SCI patients, with a purposive sub-sample of 30 MRI studies for model generation. A sample of 30 3D-printed SCI models will be produced, encompassing three lesion archetypes (compression, contusion, transection) with varying severities, replicated thrice to assess consistency. Data collection instruments include (a) 3D imaging processing software for geometrical accuracy metrics (root-mean-square error, Dice similarity coefficient), (b) mechanical testing apparatus for indentation and flexion-extension stiffness, (c) spectrophotometric methods for surrogate neural tissue proxies’ optical properties, and (d) structured performance checklists and Likert-scale questionnaires for educational utility. Validity and reliability will be established through pilot testing, inter-rater reliability in geometric measurements (ICC > 0.85), and calibration of mechanical tests against cadaveric benchmarks (within 10% error). Data analysis employs descriptive statistics and inferential tests, including repeated-measures ANOVA to compare mechanical fidelity across lesion types, Bland-Altman analysis for model-cadaveric agreement, and regression analyses to explore predictors of educational gain. A thematic analysis, guided by grounded theory, will synthesize qualitative interview data from participants, with coding conducted by two independent researchers and a third arbitrating discrepancies. The anticipated findings include high geometric fidelity of patient-specific models with Dice coefficients above 0.88 and RMS surface deviations under 0.9 mm, coupled with mechanical properties that approximate native tissue ranges (spinal cord surrogate modulus within 20% of literature values; vertebral and disc analogs within 15%). It is expected that lesion modularity will yield reproducible injury phenotypes, enabling standardized rehearsal of surgical decompression, stabilization, and regenerative interventions. Educational evaluations are predicted to show statistically significant improvements in task performance (p < 0.05) and reductions in intrinsic cognitive load (p < 0.05) after model-assisted training, with qualitative data highlighting improved spatial comprehension, procedural planning, and confidence in clinical decision-making. The study contributes to knowledge by operationalizing a scalable, end-to-end pipeline for patient-specific SCI modeling that integrates imaging, materials science, biomechanics, and medical education. It provides a validated framework for translating individualized anatomical data into tactile, evalutable simulators that can inform preoperative planning, surgeon training, and device testing under ethically and logistically feasible conditions. The models enable controlled investigations into injury mechanisms, therapeutic strategies, and the impact of anatomical variability on surgical outcomes, thereby bridging gaps between computational simulations, cadaveric studies, and clinical practice. The main conclusion is that patient-specific 3D-printed SCI models can achieve high anatomical and biomechanical fidelity while delivering measurable educational benefits, supporting broader adoption in surgical planning and research. Recommendations include expanding the repertoire of lesion types, integrating real-time haptic feedback, exploring patient-specific models for preoperative simulation in complex deformities, and establishing multi-institutional repositories to standardize model generation protocols and data sharing for reproducibility.