Skip to content

MIPT-Oulu/3DHistoGrading

Repository files navigation

3DHistoGrading

Software for 3D grading osteochondral Phosphotungstic acid -stained tissue samples. We currently support only Windows.

Build status DOI

Analysis pipeline

Background

This repository contains a software prototype and training codes used to assess degenerative features of osteochondral samples. Samples should be imaged with micro-computed tomography using Phosphotungstic acid stain. Detailed describtion for imaging and grading procedure can be found from our previous paper:

Nieminen HJ, Gahunia HK, Pritzker KPH, et al. 3D histopathological grading of osteochondral tissue using contrast-enhanced micro-computed tomography. Osteoarthritis Cartilage. 2017;25(10):1680-1689.

The texture analysis methods used for feature extraction are implemented using our LocalBinaryPattern repository.

More about the analysis procedure implemented in this repository can be found in the article published on Osteoarthritis and Cartilage.

Prerequisites

  • To avoid memory issues, the software runs on 64-bit systems only.
  • Until installer file is released, software has to be executed through MS Visual Studio. However, the compiled application can be provided upon request.

Installation

  • Currently, software can be used by opening 3DHistoGrading.sln on MS Visual Studio, compiling and running the project.
  • We are planning to create an installer file that allows installing the software without additional dependencies.

Application usage

Currently available features:

  • Visualize 3D image datasets (.png, .tiff, .bmp) using 3D rendering and three orthogonal planes
  • Load Mask on top of visualized dataset (mask should be registered with the dataset)
  • Surface artefact cropping tool for coronal and sagittal plane
  • Automatic sample alignment
  • Automatic segmentation of bone-cartilage -interface using CNTK (Microsoft Cognitive Toolkit)
  • Automatic bone (calcified tissues) and articular cartilage segmentation
  • Automatic extraction of different volumes-of-interest (surface cartilage, deep cartilage and calcified tissue)
  • Automatic grading from different osteochondral zones

Outputs

  • Result of degeneration detection (logistic regression) and corresponding µCT grade (ridge regression) from analysed sample volumes-of-interest
  • Extracted analysis volumes can be saved as separate datasets
  • Automatically segmented calcified tissue mask can be saved
  • Sample data with performed processing steps

Examples

Screenshots from our software in action

License

This software is distributed under the MIT License. This software and the pretrained models can be used only for research purposes.

Citation

If you use the software or the source code in your work, please cite our paper.

@article {Rytky713800,
		title = {Automating Three-dimensional Osteoarthritis Histopathological Grading of Human Osteochondral Tissue using Machine Learning on Contrast-Enhanced Micro-Computed Tomography},
 author = {Rytky, S.J.O. and Tiulpin, A. and Frondelius, T. and Finnil{\"a}, M.A.J. and Karhula, S.S. and Leino, J. and Pritzker, K.P.H. and Valkealahti, M. and Lehenkari, P. and Joukainen, A. and Kr{\"o}ger, H. and Nieminen, H.J. and Saarakkala, S.},
 journal = {Osteoarthritis and Cartilage}
 year = {2020},
}