160 - Improving Learning of Complex 3D Neuroanatomical Structures with 3D Labeling

Poster Board Number: 160
There are separate poster presentation times for odd and even posters.
Odd poster #s – first hour
Even poster #s – second hour

Abstract Body : Introduction and Objectives
Learning neuroanatomical structures with complex 3D surfaces is challenging. Previous studies have shown that this learning requires increased cognitive effort, spatial visualization and skill at mental rotation especially when 2D learning materials are used for study. Recent improvements in 3D scanning and printing have resulted in much better access to virtual or physical models of anatomical structures outside of lab. But often these models are unlabeled or indirectly labeled with 2D guides. This study will compare virtual 3D models of the ventricles that are unlabeled with models labeled in 3D.

Materials and Methods
A model of the human ventricles used in lab was 3D scanned using photogrammetry and labeled using Microsoft 3D Paint® and distributed as a 3D image in a Microsoft Word® document. The participants were 24 third year physical therapy students who completed the neuroanatomy course the previous year. The students were randomly assigned into two equal groups matched on gender and prior neuroscience grades. One group received an unlabeled model and the other the labeled model. Both groups had 36 hours to study the models and additional materials if desired. All students then took a 12-question matching quiz to test their ability to identify ventricular structures in a 2-D lateral view of the ventricles. They also completed a qualitative Likert style survey on the use and effectiveness of the models.

Results
All students completed the quiz and survey. Mean scores for each group were calculated.
The mean score for the labeled model quiz was higher than the unlabeled model score (91.7±3.2; 80.3±6.3 SEMs) but not significantly higher (p=0.12, 2-sample t-test).
Examining the data suggested that both groups did well on questions that are more difficult, the overall mean was used to divide the questions into easy and difficult subgroups for analysis with a mixed repeated measures ANOVA design. This analysis showed that the labeled model vs. unlabeled model scores did differ significantly (F=4.12; DF 1,44; p=0.048) with means of 86.7 vs. 65 on difficult questions and 95.2 vs 90.5 on easier questions.

The qualitative result survey found no significant difference between groups on instructions, adequate time to study, ease of use and helpful study tool. But when asked if the study material was in conjunction with the quiz. The unlabeled median was higher 4.1 (Agree) vs labeled median 3.3 (Neutral) (Mann-Whitney p=0.39). Possibly because The participants with the unlabeled model felt that it matched the unlabeled quiz image better than the participants in the labeled model group.

Conclusion
There was better learning with the labeled group. The participants with the unlabeled model felt that it matched the assessment material better than the participants in the labeled model. There was valuable feedback on improvements for the next study.

Significance/Implication
The result suggests that virtual 3D models with integrated labeling helps students improve their ability to recognize complex 3D structures in 2D images. In future studies will look at the effect of labeling on virtual 3D assessments and on lab models and specimens.