Summer ITKv4 ClinicalGroupMeetingNotes
Lesion Sizing Toolkit
Inputs: dicom Outputs: measures of volumes, segmentations
- Already in ITK as an external module. Contract is to reimplement with ITKv4 and distribute.
- Using spatial objects as inputs and outputs
- Should some of these algorithms be migrated into ITK proper? (e.g. enhanced canny edge detection)
What support it needed?
- Does ITK want a tighter integration of these classes, and in this same form? Does this cover more general concepts useful to other groups.
- Representing measures as a concept in ITK
Data: 60 datasets. Chest CT scans 1mm resolution. 200mb each. MIDAS? Store as DICOM? Automatically download using CTest.
- 3D Real-Time Physics-Based Non-Rigid Registration for Image Guided Neurosurgery (PBMNRRegistration)
The following is a rough pipeline of the method with proposed classes. (segmentation -> mesh gen -> registration -> petsc -> generate image (deformation field)) Inputs: a mask, a mesh
1) FeaturePointSelection3dFilter(mask, fixed image): start with this filter 2) BlockMatching3Dfilter(fixed, moving, metrics,feature points) -- GPU, similar to penn fem registration classes, maybe metric plug in 3) PBMSolver(displacement vector, mesh) --petsc dependence 4) ImageWarp--already in ITK
Outputs: deformation field, transformed image(s)
- Mesh generation, tetmesh reader / converter: Biomesh3D and bridge to ITK?
- ITK Mesh
- Self-updating transform object
- Integration of solver
- MPI with ITK?
What support is needed?
- CMake integration w/ petsc (uses MPI). Build / distribution
- Interface w/ FEM guys, registration & GPU guys
- ITK Algorithms for Analyzing Time-Varying Shape with Application to Longitudinal Heart Modeling
- 25 longitudinal cardiac DE-MRI (1.25mm in-plane, 2.5mm thick) with segmentations of the left atrium. 2-4 datapoints each (pre ablation, 3mo, 6mo, 1 year)
- Need IRB to release image data
- Multivariate stats.
- Bridge to R for complex statistical analysis without going to file system.