Release Notes 6.3

[iyerkrithika21]

Please add to this issue a paragraph (plus videos/snapshots for illustration) about each new/improved/bug-fix included in this release.

[jadie1]

Did we mean to close this?

Here are some release notes:

Use Cases Refactoring

• Added grooming steps to mesh-based use cases using the mesh Python API
• Alignment transformations are now passed to the optimizer and used in optimization instead of being applied before optimization. This results in local particles in the original data's coordinate system, allowing for easier subsequent analysis.
• The use cases now use project spreadsheets in optimizations instead of XML files. This format is more interpretable and allows of better integration with Studio. The project sheets support multiple domains, fixed domains, constraints
• The femur use case has been refactored into a single use case where alignment transforms and cutting plane constraints are passed in optimization.
• Grooming added for multiple domain use cases. The pipeline demonstrates alignment w.r.t domain 1 ellipsoids.
• All use case documentation has been updated to be clearer and reflect these changes.

DeepSSM Use Case

• The DeepSSM use case has been updated to demonstrate the full pipeline, including training data generation instead of relying on the femur use case to create a training shape model.
• The use case now demonstrates how to optimize validation particles via fixed domain optimization where the training particles are unchanged.
• Image-to-image registration tools have been added to prepare DeepSSM input images without requiring corresponding segmentations or meshes. This allows for true inference with DeepSSM.
• Automatic verification has been added for the DeepSSM tiny-test.
• Detailed documentation has been added to explain each step of the full DeepSSM pipeline.

Incremental Use Case

• A new use case has been added, demonstrating how a shape model can be optimized incrementally on 3D supershapes. This approach is beneficial when the cohort of shapes is very large, and single optimization would be slow, and when the dataset is small but contains a large amount of shape variation.
• Functionality has been added to select the order of shape optimization based on the distance of each shape to all others in the cohort. This allows for particles to be fit to inlier shapes first, then outliers.
• Documentation has been added that explains the use case and quantitatively demonstrates the benefit of incremental optimization.

[HeavenlyBerserker]

Constraints

• Added constraints functionality for the mesh domain both clipping and augmented lagrangian together with a flag to flip between the two options.
• Added documentation for using constraints in Studio.
• Added a Studio use case for constraints and a pseudo-tutorial for it in the documentation.

[iyerkrithika21]

Femur: Group Difference Statistics in Python was updated to perform LDA

• The use case also demonstrates Linear Discrimination of Variation (LDA) for analyzing shape variation between the subgroups.

[akenmorris]

• Color maps

Added new support for selecting different color maps. Added ability to reverse colormaps and use either discrete colors or continuous.

• Difference to mean

Added new support for showing the difference to the mean for any given mesh (subjects or generated PCA mode positions):

• Superimpose and surface distance

Added new support for displaying multiple image types (e.g. original vs groomed) with individual opacity settings. Also ability to show surface to surface distance.

• Ability to apply groom to all domains

The multiple domains groom section has been upgraded to its own panel for better UI consistency. A new checkbox for applying grooming steps to all domains has been added to assist when many domains are in use:

• Image export

A new image export dialog as been added with various options:

• PCA Montage and Fringe plot export:

Building on the image export dialog, the PCA Montage exporter allows you to create a multi-image montage across PCA modes.

• Export scalar values

Addition export options have been added to export mesh scalars, particle scalars, and all subjects particle scalars.

• Image volume support

New support has been added for displaying 2D slices from image volumes (e.g. CT/MRI):

• Landmark placement UI in Studio, landmark registration, landmarks as initializers

video:

https://sci.utah.edu/~shapeworks/doc-resources/mp4s/studio_landmarks.mp4

• Cutting planes UI in Studio

Added new support for defining and manipulating cutting planes

• Free form constraints UI in Studio

Added new support for defining free form constraints

• Procrustes scaling only mode

New support for running procrustes in a scaling-only mode has been added.

• Good/bad particle display

The Particles Panel enabled the display of "good/bad particles" in ShapeWorks Studio. A maximum angle parameter is given (default 45 degrees). When a particle's surface normal across subjects (averaged) exceeds this angle, the particle will be marked "bad" and shown in red. Good particles are shown in green. This can help in identifying problems with correspondence as it is typically expected that surface normals for given particles should be similar across subjects.

• Group LDA chart in Studio

Support for the group LDA chart has been added in Studio:

[akenmorris]

Complete in #1777