Medical Imaging Segmentation & 3D Model Refinement
Budget: $30 – $250 USD
Project 2: Medical Image Segmentation and 3D Geometry Smoothing
Project Description
3D SLICER AND FUSTION
In this project, you will segment and smooth a feature from a medical imaging dataset. You may either use one of 3D Slicer’s Sample Data sets (File → Download Sample Data) or obtain a dataset from an external source. Example public datasets include the FastMRI DatasetLinks to an external site. (knee, brain, prostate, and breast scans from NYU) and the MRNet Knee MRI DatasetLinks to an external site. from Stanford.
https://fastmri.med.nyu.edu/
https://aimi.stanford.edu/datasets/mrnet-knee-mris
From the medical image you choose, select one anatomical structure to segment. Examples include a bone or cartilage from a knee MRI, a tumor or skull from the MRBrainTumor sample dataset, or an organ from the CTLiver sample dataset.
When selecting a structure, choose one that is clearly visible and distinguishable in the scan. Structures with strong contrast (such as bones in CT scans) are typically much easier to segment than soft tissues, so consider segmentation difficulty when making your selection.
If you are unsure whether your chosen structure is appropriate, please check with the instructor before beginning.
You will segment the selected structure in 3D Slicer to generate a 3D model. After segmentation, export the model and import it into Fusion 360 to smooth and clean up the geometry (for example, smoothing surfaces, removing artifacts, or repairing mesh issues) to produce a refined 3D model.
Report Requirements (2–3 Pages)
Prepare a short report describing your work. Your report should include the following sections.
Introduction
Describe the imaging dataset you used. Indicate whether it came from 3D Slicer’s Sample Data or an external source. Explain which anatomical structure you chose to segment and smooth, and why you selected that particular scan and structure. Include references for any outside datasets or sources used.
Methods
Explain the steps you followed during the segmentation and smoothing process. Describe your workflow in both 3D Slicer and Fusion 360. Include screenshots where helpful to illustrate key steps in your process. Provide relevant technical details such as thresholding values, segmentation tools used, or smoothing parameters.
Results
Present the final smoothed 3D geometry you created. Include images of the model and quantitative metrics describing it. Use Fusion 360 to estimate properties such as volume and mass (you may assume a reasonable material density if needed).
Discussion
Reflect on your process. Were any parts of the segmentation or smoothing particularly challenging? How well does the final geometry represent the structure you intended to capture? Does the estimated mass seem reasonable based on what you know about the structure? Explain your reasoning.
Submission Requirements
Submit a ZIP folder containing the following files:
The original scan data used for your project (if using an external dataset). Include only the files for the single scan you used (typically the DICOM slices for that scan).
3D Slicer Scene file (.mrml)
Segmentation file (.seg.nrrd)
Cleaned 3D geometry (.stl & .step)
Project report (Word document)
Grading Rubric (100 pts)
Introduction – 5 pts
Dataset and source clearly described
Structure being segmented identified and justified
Methods Description – 20 pts
Clear explanation of segmentation and smoothing steps
Technical details included (tools, thresholds, parameters)
Screenshots effectively show the workflow
Segmentation Quality – 30 pts
Structure correctly segmented
Segmentation captures the intended anatomy
Minimal artifacts or major errors
Geometry Cleanup / Fusion 360 Work – 20 pts
Model exported and processed correctly
Geometry smoothed and cleaned appropriately
Final STL and STEP are usable
Results and Analysis – 15 pts
Final model clearly presented
Metrics reported (volume, estimated mass)
Results interpreted appropriately
Discussion – 5 pts
Thoughtful reflection on challenges and outcomes
Organization and Clarity – 5 pts
Report is clear, organized, and follows required sections
Project Description
3D SLICER AND FUSTION
In this project, you will segment and smooth a feature from a medical imaging dataset. You may either use one of 3D Slicer’s Sample Data sets (File → Download Sample Data) or obtain a dataset from an external source. Example public datasets include the FastMRI DatasetLinks to an external site. (knee, brain, prostate, and breast scans from NYU) and the MRNet Knee MRI DatasetLinks to an external site. from Stanford.
https://fastmri.med.nyu.edu/
https://aimi.stanford.edu/datasets/mrnet-knee-mris
From the medical image you choose, select one anatomical structure to segment. Examples include a bone or cartilage from a knee MRI, a tumor or skull from the MRBrainTumor sample dataset, or an organ from the CTLiver sample dataset.
When selecting a structure, choose one that is clearly visible and distinguishable in the scan. Structures with strong contrast (such as bones in CT scans) are typically much easier to segment than soft tissues, so consider segmentation difficulty when making your selection.
If you are unsure whether your chosen structure is appropriate, please check with the instructor before beginning.
You will segment the selected structure in 3D Slicer to generate a 3D model. After segmentation, export the model and import it into Fusion 360 to smooth and clean up the geometry (for example, smoothing surfaces, removing artifacts, or repairing mesh issues) to produce a refined 3D model.
Report Requirements (2–3 Pages)
Prepare a short report describing your work. Your report should include the following sections.
Introduction
Describe the imaging dataset you used. Indicate whether it came from 3D Slicer’s Sample Data or an external source. Explain which anatomical structure you chose to segment and smooth, and why you selected that particular scan and structure. Include references for any outside datasets or sources used.
Methods
Explain the steps you followed during the segmentation and smoothing process. Describe your workflow in both 3D Slicer and Fusion 360. Include screenshots where helpful to illustrate key steps in your process. Provide relevant technical details such as thresholding values, segmentation tools used, or smoothing parameters.
Results
Present the final smoothed 3D geometry you created. Include images of the model and quantitative metrics describing it. Use Fusion 360 to estimate properties such as volume and mass (you may assume a reasonable material density if needed).
Discussion
Reflect on your process. Were any parts of the segmentation or smoothing particularly challenging? How well does the final geometry represent the structure you intended to capture? Does the estimated mass seem reasonable based on what you know about the structure? Explain your reasoning.
Submission Requirements
Submit a ZIP folder containing the following files:
The original scan data used for your project (if using an external dataset). Include only the files for the single scan you used (typically the DICOM slices for that scan).
3D Slicer Scene file (.mrml)
Segmentation file (.seg.nrrd)
Cleaned 3D geometry (.stl & .step)
Project report (Word document)
Grading Rubric (100 pts)
Introduction – 5 pts
Dataset and source clearly described
Structure being segmented identified and justified
Methods Description – 20 pts
Clear explanation of segmentation and smoothing steps
Technical details included (tools, thresholds, parameters)
Screenshots effectively show the workflow
Segmentation Quality – 30 pts
Structure correctly segmented
Segmentation captures the intended anatomy
Minimal artifacts or major errors
Geometry Cleanup / Fusion 360 Work – 20 pts
Model exported and processed correctly
Geometry smoothed and cleaned appropriately
Final STL and STEP are usable
Results and Analysis – 15 pts
Final model clearly presented
Metrics reported (volume, estimated mass)
Results interpreted appropriately
Discussion – 5 pts
Thoughtful reflection on challenges and outcomes
Organization and Clarity – 5 pts
Report is clear, organized, and follows required sections