explore: automated rib cage morphometry

[msrepo]

Description

what aspects of the rib cage mesh can be automatically measured?

Possible references: 3D reconstruction of the ribs from lateral and frontal X-rays in comparison to 3D CT-scan reconstruction

Dansereau, J., Stokes, I.A., 1988. Measurements of the three-dimensional shape of the rib cage. J. Biomech. 21 (11), 893–901

[bishesh]

Topology? How well individual ribs are separated and are not wrongly fractured/broken?

[msrepo]

Start with Groundtruth segmentation: Obviously, we can do manual endpoint marking to obtain rib centerlines and then obtain required metrics. How far can we automate this? also, caveat: vmtkCenterLine tool requires each sub-rib to be processed individually rather than in batch, which is cumbersome. refer: https://discourse.itk.org/t/centerline-extraction/4885/8

[msrepo]

How are the clinical metrics useful? Answer: https://www.sciencedirect.com/science/article/abs/pii/S0021929007004241

[msrepo]

• [ ] Collect endpoints for 73 rib test set
• [ ] tool to store and visualize these endpoints and corresponding rib mesh
• [ ] adapt tool for centerline extraction from vmtk slicer python script
• [ ] Extract metrics from the centerline

[msrepo]

there will be issues in extracting centerline if there are segmentation issues there is a need to postprocess even groundtruth segmentation some segmentations do not look like anatomically valid ribs. Lets keep in mind how many of these cases we come across.

[msrepo]

fit a polynomial curve to the centerline sometimes you do not get a centerline curve but do get a network model. why is that? case in point: Endpoints 3 Final result

[msrepo]

need to define the order in which the endpoints are picked. We will later need to compare the groundtruth with the model prediction at which point we do not want to solve the corresponding point problem.

[msrepo]

• visualize points stored in the Slicer format orientation needs to be fixed