Slicer3:UIDesign:WorkingProblems:BCAnalysis:Volumetric
From Slicer Wiki
Home < Slicer3:UIDesign:WorkingProblems:BCAnalysis:Volumetric
Back to Project Overview
Contents
Volumetric Tumor Analysis for Breast Cancer
Goal
Goal is to accommodate basic RECIST assessment for breast cancer (Response Evaluation Criteria In Solid Tumors). (RECIST is a set of published rules that define when cancer patients improve ("respond"), stay the same ("stable") or worsen ("progression") during treatments).
- Determine features that currently exist in Slicer
- Develop features that don't
- Provide a framework that knits features together in a comfortable workflow.
Linear Measurements Workflow Described by Jeffrey Yapp
- Scroll through baseline
- Find the tumor's longest diameter (no reformat, just on orig acquisition)
- Measure the longest diameter D1(t1)
- Measure the orthogonal diameter D2(t1)
- If there are multiple tumors, measure both diams in each tumor and sum to get relative D1(t1) and relative D2(t1)
- Perform same measurements D1(t2) and D2(t2)
- Specific results to report:
- Percent change
- Absolute change
- Question: units for these values -- mm!
Approach in Slicer
- create a subvolume of target breast for both pre and post treatment timepoints: try to make these two subvolumes match as closely as possible. Ron uses 2-fold oversampling.
- Registration:
- create transform node
- perform manual,
- then affine,
- then B-spline (modified number of histogram bins to 20)
- run bias field correction on both subvolumes with 10 iterations and skin mask for pre and post separately
- mask out background
- perform histogram matching with "threshold at mean" checked
- Ran diffeomorphic demons on post, took 3 hours, with nice results
- Linear Measurements:
- scroll to slice with largest diameter Dmax
- measure Dmax and orthogonal Dmax on both treatment timepoints using the linear measurement widget.
- in Editor module:
- use threshold paint to segment tumor; or
- use fastmarching with seeds in the tumor tissue
- erase any outlying pixels
- in Fiducials module:
- create two fiducials along longest axis
- check distance measure between; or
- eventually use nicole's measurement widget
- move fiducials to mark orthogonal axis
- check distance measure between;
- Extra things to do:
- use the editor quick model to create a model of the gland
- use the editor to segment the skin (threshold, change island, quick model)
Ron's requests:
- The fiducial text is too large in the fast marching display. The default text size be smaller?
- I was not quite clear on how to create a new label map for doing the skin segmentation. I ended up reusing the fastmarchinglabel and losing that segmentation.
- quickmodel should use the label name as default.
- It would be nice to have short cuts to the add label tools and label statistics tools on the editor page. This would give a cue for beginners where to look for that functionality.
Volume Measurements Workflow
- Scroll through dataset and determine tumor with largest diameter.
- Segment largest diameter tumor in baseline Question: multiple tumors ok?
- Manual or automatic segmentation OK
- Segment tumor in followup Questions: Must be same tumor, or largest diameter in followup? multiple tumors ok? (same tumor)
- Manual or automatic segmentation OK -- but use same method
- Measure:
- Volume(t1) and Volume(t2) Question: which is preferred, mm^3!
- Relative % difference
- Absolute change (same units as above)
STATUS
- have new Pre/Post GAD Ax case for post-treatment -- need to regenerate slides and workflow with this dataset instead of T1 for better tumor contrast (November RSNA deadline)
- need to use dataset appropriate for ChangeTracker demonstration -- t2 for this dataset has no detectable tumor (November RSNA deadline)