pyKNEEr: An image analysis workflow for open and reproducible research on femoral knee cartilage

Abstract

Transparent research in musculoskeletal imaging is fundamental to reliably investigate diseases such as knee osteoarthritis (OA), a chronic disease impairing femoral knee cartilage. To study cartilage degeneration, researchers have developed algorithms to segment femoral knee cartilage from magnetic resonance (MR) images and to measure cartilage morphology and relaxometry. The majority of these algorithms are not publicly available or require advanced programming skills to be compiled and run. However, to accelerate discoveries and findings, it is crucial to have open and reproducible workflows. We present pyKNEEr, a framework for open and reproducible research on femoral knee cartilage from MR images. pyKNEEr is written in python, uses Jupyter notebook as a user interface, and is available on GitHub with a GNU GPLv3 license. It is composed of three modules: 1) image preprocessing to standardize spatial and intensity characteristics, 2) femoral knee cartilage segmentation for intersubject, multimodal, and longitudinal acquisitions, and 3) analysis of cartilage morphology and relaxometry. Each module contains one or more Jupyter notebooks with narrative, code, visualizations, and dependencies to reproduce computational environments. pyKNEEr facilitates transparent image-based research of femoral knee cartilage because of its ease of installation and use, and its versatility for publication and sharing among researchers. Finally, due to its modular structure, pyKNEEr favors code extension and algorithm comparison. We tested our reproducible workflows with experiments that also constitute an example of transparent research with pyKNEEr. We provide links to executed notebooks and executable environments for immediate reproducibility of our findings. IntroductionOpen science and computational reproducibility are recent movements in the scientific community that aim to promote and encourage transparent research. They are supported by national and international funding agencies, such as the United States National Institutes of Health (NIH) [9] and the European Commission [10]. Open science refers to the free availability of data, software, and methods developed by researchers with the aim to share knowledge and tools [75]. Computational reproducibility is the ability of researchers to duplicate the results of a previous study, using the same data, software, and methods used by the original authors [5]. Openness and reproducibility are essential to researchers to assess the accuracy of scientific claims [55], build on the work of other scientists with confidence and efficiency (i.e. without "reinventing the wheel") [54], and collaborate to improve and expand robust scientific workflows to accelerate scientific discoveries [53,14,42]. Historically, research data, tools, and processes were rarely openly available because of limited storage and computational power [42]. Nowadays, there are several opportunities to conduct transparent research: data repositories (e.g. Zenodo and FigShare), code repositories (e.g. GitHub, Git...

3D MRI in Osteoarthritis

“…2 shows an example of segmented femoral cartilage based on a 3D dual-echo steady state (DESS) sequence using an automated segmentation algorithm. 7 In the 2019 International Workshop on Osteoarthritis Imaging (IWOAI) knee segmentation challenge, a standardized partition of the 3D DESS data from the Osteoarthritis Initiative was used to compare six different segmentation algorithms. 8 The highest Dice similarity coefficients (DSCs), a common metric to assess segmentation performance, for segmentation of femoral, tibial, and patellar cartilage were 0.90, 0.89, and 0.86, respectively.…”

Use of Computational Modeling to Study Joint Degeneration: A Review

“…In addition, and comparable to our discussion on musculoskeletal modeling, manual segmentation of the geometry from imaging data to build a patient-specific model is a time consuming task, and hence is not suitable to be integrated in a routine manner in the clinic. There have been quite a few approaches to automate the process of segmentation (Marstal et al, 2011;Lee et al, 2014;Dam et al, 2015;Ye et al, 2015;Bonaretti et al, 2019). However, most approaches are directed at automatic segmentation of the articular cartilage, while omitting the other significant tissues of the joint such as the ligaments or the meniscus [except for Dam et al (2015), where meniscus was included], which will restrict the accuracy of the in silico models to predict the in vivo conditions.…”