Quasi-automated reconstruction of the femur from bi-planar X-rays

ABSTRACT

3D reconstruction from low-dose Bi-Planar X-Rays (BPXR) is a rising practice in clinical routine. However, this process is time consuming and highly depends on the user. This  study aims to partially automate the process for the femur, thus decreasing reconstruction time and increasing robustness. As a training set, 50 femurs are segmented from CT scans together with 120 BPXR reconstructions. From this dataset, an initial solution for the bony contours is defined through Gaussian Process Regression (GPR), using eight digitized landmarks. This initial solution is projected on both x-rays and automatically adjusted using an adapted Minimal Path Algorithm (MPA). To evaluate this method, CT-scans were acquired from 20 cadaveric femurs. For each sample, the CT-based reconstruction is compared to the one automatically generated from the digitally reconstructed radiographs. Euclidean distances between femur reconstructions and the segmented CT data are on average 1.0 mm with a Root Mean Square Error (RMSE) of 0.8 mm. Femoral torsion errors are assessed: the bias is lower than 0.1° with a 95% confidence interval of 4.8°. The proposed method substantially improves 3D reconstructions from BPXR, as it enables a fast and reliable reconstruction, without the need for manual adjustments, which is essential in clinical routine.

KEYWORDS:

• Biomechanics
• biomedical Imaging
• computer Methods
• femur
• bi-planar X-rays
• 3d reconstruction

Acknowledgments

We would like to thank Maxim Van den Abbeele and Bhrigu Lakhar for their critical comments, which greatly improved the quality of this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

François Girinon

François Girinon received his master degree in Biomedical Enginering from Arts et ParisTech in 2015 and the PhD degree in Biomedical imaging from Arts et Métiers ParisTech in 2018. His general interests lie in geometric modeling and medical image analysis for the 3D reconstruction of the lower limbs.

Laurent Gajny

Laurent Gajny is assistant professor in applied mathematics at Arts et Métiers ParisTech. His research interests are in numerical analysis and geometric modelling applied to the 3D reconstruction of the human body from medical images.

Shahin Ebrahimi

Shahin Ebrahimi received his master degree in Biomedical Enginering from Telecom ParisTech in 2014 and the PhD degree in Computer Science from Arts et Métiers ParisTech in 2017. His general interests lie in machine learning, pattern recognition and their application to computer vision and medical image analysis.

Louis Dagneaux

Senior consultant and Assistant Professor in Orthopedic surgery (Montpellier, France) for five years, Louis Dagneaux is currently working at the Mayo Clinic as visiting scientist. With the support of over 25 peer-reviewed publications and 3 book chapters, he focuses his research and clinical practice on patellofemoral factors related to knee arthroplasty and arthroscopic foot and ankle surgery using mechanics of the lower-limb.

Philippe Rouch

Professor Philippe Rouch is Paris Campus Director of the Arts et Métiers ParisTech. He has also been director of the Institut de Biomécanique humaine Georges Charpak in Art et Métiers ParisTech. He is particularly involved in musculoskeletal modeling with a strong interest in kinematic analysis.

Wafa Skalli

Wafa Skalli is a professor in biomechanics at Arts et Métiers ParisTech. She is founder and scientific director of the Institut de Biomécanique Humaine Georges Charpak in Arts et Métiers ParisTech, and holder of the BiomecAM  ParisTech chair on subject-specific musculoskeletal modelling. She is particularly involved in biomechanics and modelling of the spine, with a strong link to experimental and clinical approach.