constructing a Statistical Shape Model (SSM) of the femoral cartilage

1

Hello everyone,

I would like to attempt constructing a Statistical Shape Model (SSM) of the femoral cartilage using Shapeworks. I’ve encountered some issues along the way and I’m wondering if anyone has experience in doing something similar. Your insights would be greatly appreciated. Thank you!

2

I’m not sure if anyone has modeled femoral cartilage. Can you describe the issues you are encountering?

3

I have used your shapeworks Python API for the construction of a cartilage Statistical Shape Model (SSM). However, due to a significant amount of noise in the data (resulting in cartilage erosion and suboptimal segmentation), the constructed cartilage SSM model is not as ideal as desired. The results of the construction are as shown in the attached image.

image.png
image.png5760×2241 2.07 MB

Alan Morris via ShapeWorks <notifications@shapeworks.discoursemail.com> 于2023年9月6日周三 02:57写道：

4

Yes, this seems to be a data preprocessing/grooming issue. I would not proceed with optimization until we can resolve the preprocessing.

Are your original shapes coming from image segmentations / labelmaps or are they starting as meshes? Could you possibly share 1-2 of them? amorris@sci.utah.edu

5

I have utilized the SKI-10 challenge dataset (https://ski10.grand-challenge.org/) for knee joint research. I have extracted label masks for knee cartilage and directly converted them into meshes. I will be looking forward to receiving the mesh files you have used. “Could you please check if this kind of data meets your data processing requirements?”

image.png
image.png931×738 288 KB

Best
Jingyi

Alan Morris via ShapeWorks <notifications@shapeworks.discoursemail.com> 于2023年9月6日周三 23:57写道：

6

You might have better luck if you import the segmentations into ShapeWorks and let the image segmentation grooming tools work on the data before the mesh construction (which ShapeWorks will also do). This will eliminate holes, stairsteps, etc.

You might also consider modeling the bone first and then bone+cartilage.

7

I am working with segmentation results that include masks for four different labels(femur 、femur cartilage 、 tibia 、 tibia cartilage). Specifically, I have masks for the femur and femur cartilage. Should I groom these masks together or separately? I have been looking through your Python API but could not find an interface that allows for optimization of this kind of data. If there are any tutorials or resources that address this issue, could you please point me in the right direction? I would greatly appreciate your guidance. Thank you very much for your time and assistance. Best

Alan Morris via ShapeWorks <notifications@shapeworks.discoursemail.com> 于2023年9月8日周五 02:10写道：

8

I would suggest starting with the femurs alone and make sure you can create a good shape model. Then try femur+cartilage.