Segmentation of trabecular bone microdamage in Xray microCT images using a two-step deep learning method

January 24, 2023

Rodrigue Caron (1) (2), Irène Londono (2), Lama Seoud (2) (3) (4), Isabelle Villemure (1) (2) (3)
Journal of the mechanical behavior of biochemical materials, volume 137. (January 2023). DOI:


Trabecular bone, microdamage, microCT, deep learning


One of the current approaches to improve our understanding of osteoporosis is to study the development of bone microdamage under mechanical loading. The current practice for evaluating bone microdamage is to quantify damage volume from images of bone samples stained with a contrast agent, often composed of toxic heavy metals and requiring long tissue preparation. This work aims to evaluate the potential of linear microcracks detection and segmentation in trabecular bone samples using well-known deep learning models, namely YOLOv4 and Unet, applied on microCT images.

How Our Software Was Used

Outputs (reference bounding boxes) from Dragonfly’s YOLO deep learning architecture were used as masks for the training and testing of a U-Net deep model. This made training much faster and provided the opportunity to use more data augmentation.

Author Affiliation

(1) Department of Mechanical Engineering, Polytechnique Montréal, Montréal, QC, Canada
(2) Centre de recherche du CHU Sainte Justine, CHU Sainte Justine, Montréal, QC, Canada
(3) Institut de génie biomédical, Montréal, QC, Canada
(4) Department of Computer Engineering and Software Engineering, Polytechnique Montréal, Montréal, QC, Canada