Neuroanatomy can be difficult to learn from flat images alone: students need to understand how brain structures relate to one another in three dimensions. Formalin-preserved specimens can also be limited in availability and have a strong odor. A study at the University of Sarajevo’s Veterinary Faculty examined whether color-coded 3D-printed horse brain models could support teaching and how students responded to using them.
Key findings
- Students had 15 hours of teaching with 3D-printed models and the remaining 3 hours with formalin-preserved specimens.
- The teaching sessions used 32 models, providing a one-to-one student-to-specimen ratio.
- For 15 anatomical structures, identification accuracy ranged from 59.45% to 100% on 3D models and from 56.75% to 94.59% on formalin-preserved specimens.
- A paired-samples test found a difference favoring the 3D models: t(14) = 4.80, p = 0.00023.
- Students rated color-coding highly, with a mean satisfaction score of 4.76 ± 0.43.
- The experimental cohort achieved better neuroanatomy quiz results than two earlier cohorts taught with formalin-preserved specimens.
Why this question matters
Anatomy underpins veterinary education because understanding the location and relationships of body structures supports the study of physiology, disease, and clinical practice. Neuroanatomy is particularly demanding: learners must recognize brain regions and understand their spatial relationships. Teaching has traditionally relied on cadavers and preserved specimens, but maintaining and accessing them can be challenging. Formalin also has a strong odor, and students in the study described concerns about possible health effects.
3D printing, also known as additive manufacturing, makes it possible to create physical teaching aids from digital models. A printed model can be reused, and color can help distinguish individual structures. However, a printed object is not biological tissue. It is therefore important to assess whether it supports learning and identification rather than assume that it can fully replace specimens.
What the study tested
The study compared first-year students in the integrated veterinary study program during the 2022/2023 academic year with two earlier cohorts, from 2021/2022 and 2020/2021. The 2022/2023 cohort was the experimental group. Earlier cohorts studied neuroanatomy exclusively with formalin-preserved specimens. In the experimental group, 15 of the 18 course hours used 3D models, while formalin-preserved specimens were used for the other three hours. Teaching was consistent across all three cohorts, and the groups shared the same year of study and admission criteria.
The teaching materials began as two horse brain models made manually from gypsum. One showed the ventral surface and the origins of the cranial nerves, with a section through the lateral ventricles, mesencephalon, and rhomboid fossa on the opposite side. The other was a sagittal section showing the cerebral hemisphere, cerebellum, and corpus callosum.
How the models were made
The gypsum prototypes were scanned with an Artec Eva 3D scanner. 3D scanning captures an object’s shape from multiple angles. Artec Studio software was used to clean and align the scans and create an STL model. STL is a geometry file format commonly prepared for 3D printing. The files were prepared for printing in Cura 5.6.0.
FDM printing was used. FDM, or fused deposition modeling, builds an object layer by layer by extruding material through a heated nozzle. The models were printed in PLA filament on Ultimaker 3 and Raise3D Pro 3 Plus printers. They were printed vertically with support structures, using the Cura slicer’s predefined Normal settings. Supports were removed after printing, and the anatomical structures were hand-painted with oil paints. Four anatomy professors supervised the preparation and checked the models’ morphological accuracy, dimensions, and structural integrity. A total of 32 models were prepared for practical sessions.
Learning and structure-recognition results
Students’ knowledge was assessed with an online quiz on Moodle. The system randomly selected 40 questions from a bank of more than 400. Four course teachers prepared the questions, which included matching labeled structures to terms, multiple-choice questions, and true-or-false questions. The images came from standard course literature and teaching materials available to all cohorts.
The experimental cohort achieved better neuroanatomy test results than both earlier cohorts. Those values should not be inferred. The discussion also reports that the statistically higher scores in the 3D-model cohort were predominantly achieved by male students. This is a finding from this study, not evidence that 3D models benefit only one group of students.
In a separate recognition test, students in the experimental cohort identified 15 structures on both 3D models and formalin-preserved specimens. Accuracy varied by structure, but the overall reported range was higher for the printed models. The paired-samples test compares the same participants’ results on both specimen types and showed a statistically significant difference favoring the 3D models. Students gave the same responses for four structures on both types; for the others, identification accuracy was higher on the printed models.
What the findings mean for teaching and design
Students responded positively to the models and especially valued color-coding as a way to distinguish anatomical regions. The mean rating for color-coding was 4.76 ± 0.43. Students also described the models as making course content easier to navigate. Those with chronic respiratory problems or allergies reported feeling safer with the 3D models, citing the odor of formalin-preserved specimens and concerns about health effects.
For educators, the results support using 3D models as additional teaching resources: they are reusable and provide a visually clear object for individual practice. For engineers and designers, the study illustrates a workflow from scanning and processing an STL file through FDM printing to hand-painting. The study did not compare different filament types, print settings, or other technologies such as SLA. It therefore cannot establish which alternative configuration would produce better educational results.
The discussion gives an estimated initial 3D-printer investment of USD 500–3,000 and a material cost of under USD 5 for one brain model, with a similar additional amount for painting services. These are estimates cited in the paper, not a full accounting of all equipment, labor, and production costs.
Limitations and next steps
One important limitation is that the comparison used two historical control cohorts rather than a concurrent control group. Although the cohorts were close in time, shared a teacher, and had consistent admission criteria, comparisons across academic years cannot rule out every other difference in student experience. The study also focused on two models. The paper notes that additional sections and models representing more brain regions could provide further teaching benefits. Future evaluations are also recommended to include anatomy educators and practitioners, rather than relying exclusively on student assessments.
Conclusion
In this study, color-coded 3D-printed horse brain models were well received and were associated with better quiz results and more accurate structure identification than earlier cohorts and formalin-preserved specimens. The measured results do not imply that printed models should replace biological material altogether. Instead, PLA models made with FDM appear to be a practical complement to veterinary neuroanatomy teaching. The study by Adi Pandžić, PhD, was published in BMC Medical Education in 2025.
| Authors | Nedžad Hadžiomerović, Rizah Avdić, Adis J. Muminović, Abdullah Muftić, Adi Pandžić, Faruk Tandir, Anel Vejzović, Ozan Gündemir, Amina Isanović Hadžiomerović |
|---|---|
| Published in | BMC Medical Education, 2025, volume 25, article 1323 |
| Publication date | October 2, 2025 |
| DOI | 10.1186/s12909-025-07908-y |
| Full paper | researchgate.net |
Frequently asked questions
Do 3D-printed models improve neuroanatomy learning?
In this study, the cohort taught with the models had better quiz results than two earlier cohorts. Because the controls were historical, the comparison does not prove that the models alone caused the entire difference.
What material and printing technology were used?
The models were printed in PLA using FDM technology on Ultimaker 3 and Raise3D Pro 3 Plus printers. They were hand-painted with oil paints after printing.
How accurately did students identify structures on the 3D models?
Across the 15 tested structures, accuracy on 3D models ranged from 59.45% to 100%. Accuracy on formalin-preserved specimens ranged from 56.75% to 94.59%.
This article summarises a published research paper; please refer to and cite the original paper. About the author
