2019/11/05 by Joeri Nicolaes, Nicolaes, Joeri, Steven Raeymaeckers +11
Engineering · Medicine · #Advanced X-ray and CT Imaging #Bone and Joint Diseases #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Medical Imaging and Analysis #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1911.01816
openalex publication_date 2019/11/05 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Osteoporosis induced fractures occur worldwide about every 3 seconds.\nVertebral compression fractures are early signs of the disease and considered\nrisk predictors for secondary osteoporotic fractures. We present a detection\nmethod to opportunistically screen spine-containing CT images for the presence\nof these vertebral fractures. Inspired by radiology practice, existing methods\nare based on 2D and 2.5D features but we present, to the best of our knowledge,\nthe first method for detecting vertebral fractures in CT using automatically\nlearned 3D feature maps. The presented method explicitly localizes these\nfractures allowing radiologists to interpret its results. We train a\nvoxel-classification 3D Convolutional Neural Network (CNN) with a training\ndatabase of 90 cases that has been semi-automatically generated using\nradiologist readings that are readily available in clinical practice. Our 3D\nmethod produces an Area Under the Curve (AUC) of 95% for patient-level fracture\ndetection and an AUC of 93% for vertebra-level fracture detection in a\nfive-fold cross-validation experiment.\n