The Use & Efficacy of Interactive Case Studies in Teaching Researchers about Pseudoprogression
Summary
Tumor pseudoprogression is defined as an increase in tumor size after immunotherapeutic treatment without true tumor growth, followed by a size reduction despite no changes in therapy. Pseudoprogression visually mimics true tumor progression in radiological scans, posing an important diagnostic challenge for physicians and researchers. If misdiagnosed, patients may be prematurely removed from treatment. Radiological imaging is the most common method of diagnosis and monitoring treatment effectiveness in solid tumor therapies but radiological scans often contain complex anatomical and spatial relationships that are difficult to visualize.
Interactive three-dimensional (3D) visualization has been shown to improve visuospatial and diagnostic ability in radiology students. Additionally, case-based learning (CBL) has been commonly used in medical education to teach diagnostic skills and complex biomedical information. Case studies integrated within an interactive 3D module could be an effective way of teaching biomedical researchers about pseudoprogression. Two web-based interactive 3D instructional tools will be created using mouse glioblastoma MRI data, one presented in a case study format and the other in a non-case study format, to evaluate the impact of interactive case studies in teaching researchers about the diagnosis of tumor pseudoprogression.
Objective:
This study aims to understand whether the presentation of the topic of pseudoprogression in the form of interactive case studies can enhance biomedical researcher’s understanding of tumor pseudoprogression and increase their comfort levels in viewing radiological scans.
Software:
Unity 3D, ZBrush, 3DSMax, Materialize Mimics, ITK-SNAP, Articulate Storyline 360, Adobe Illustrator, After Effects, XD
Research Committee:
Samantha Bond (Advisor)
Biomedical Visualization, UIC
Christine Young, Dr. Leah Lebowicz
(Committee Members)
Biomedical Visualization, UIC
Content Advisors:
Dr. Daniele Procissi
Radiology Department, Feinberg School of Medicine, Northwestern University
Dr. Karen Xie
Radiology Department, College of Medicine, UIC
Pre-Production
After conducting extensive literature review, I created a set of low and high fidelity wireframes for the interactive learning modules using Adobe Illustrator and Adobe XD.
E-learning module
Two versions of the learning module were wireframed; the case study-based learning version and non case study-based learning version. The image below shows high fidelity wireframes for the case-based version of the module.
Interactive tumor visualization platform
Wireframes for the interactive platform were carefully designed to allow incorporation of multiple types of data so that the platform can be a comprehensive visualization tool. The UI of the platform was revised multiple times to ensure smooth navigation of tumor 3D data, MRI scans and fluid biomarker data over time.
Production
All 2D illustrations and UI animations were created in Adobe Illustrator and After Effects. The following section describes the 3D to interactive workflow.
Segmentation
Mouse brain and glioblastoma MRI data were converted into DICOM format and segmented using Materialize Mimics and ITK-SNAP. Because there is low contrast difference between the three brain regions, masks were painted manually while constantly referencing the Allen Mouse Brain Connectivity Atlas to ensure accuracy is maintained. Tumors and necrotic regions were segmented using a combination of manual and semi-automated segmentation.
Model Optimization
Segmented models of the mouse brain and tumors were imported into ZBrush for further processing and smoothing. The models were smoothed out and retopologized to obtain low poly models. Models with the lowest subdivision level was exported into Unity and its corresponding normal maps and polypaint were projected onto the model to obtain a model with sufficient detail. This workflow ensured that models were well optimized for use on an interactive application hosted on a WebGL platform and built using the Unity game engine. Before importing into Unity, the tumor models were also imported into 3DSMax to create a morphing tumor from one time point to the next using morph modifiers.
Interactive Development
All models were imported into Unity in their optimized form, where they were incorporated into the interactive tumor visualization platform.
The interactive platform enabled users to view tumor morphology and biomarker data over time in a comprehensive manner. A timeline slider was linked to MRI stacks, tumor models and biomarker data graph to their respective time points. Sliding along the timeline showed the tumor and necrotic models morphing over time and the MRI stack associated with the respective time points. It also updates a sliding panel on the biomarker graph, enabling users to easily locate biomarker data along a timeline. The tumor data and other biomarkers (size, ctDNA levels, etc.) are located on an expandable panel on the right.
The following video shows a screen recording of the interactive tumor visualization platform:
The tumor visualization platform was then published on a WebGL platform and embedded in a web-based e-Learning module created using Articulate Storyline 360. In the case study styled version of the learning module, active exploration and reflection was encouraged throughout the case study through reflection and review questions. In the non case study version of the learning module, content was presented in a didactic and linear format with minimal exploration allowed and no reflection questions.
The following videos show screen recordings on how a user would progress through the case study-based module (left) compared to the non case study-based module.
Case study-based module
Non case study-based module