Technical Name Inference of bone density from X-ray using AI methods
Project Host 裴育晟
Osteoporosis is a problem that needs attention in an aging society. There are many testing methods,these methods are passive medical behaviors. When patients learn that they have osteoporosis, they are all at an irreversible level. This technology uses conventional X-ray imagesdeep learning technology to evaluate the bone structure. This method can predict bone health from low-cost X-rays,further infer whether the patient has osteoporosis. Fast, low-cost,accurate test.
Scientific Breakthrough
This technology predicts the possibility of osteoporosis through deep learning technology. The innovative process of this deep learning model enables the proposed algorithm to automatically capture the fine textures in bone structures of osteoporotic/osteopenic,thereby trains a deep learning regression model to extract those structural features to predict bone density. The performance of this technology has surpassed previous publicationshas reached the level of clinical use. Finally, a masked mode of the model is innovated to reveal the basis for model interpretation. A clinical application scheme of this interpretable artificial intelligence is proposed.
Industrial Applicability
The applicability of this technology has two directions. One is the rapid screening of osteoporosis. Through conventional X-ray examination, the possibility of osteoporosis is evaluated from X-ray images of different parts,more patients with potential loss in bone mineral density can be found. The second is to assist in tracking high-risk osteoporosis patients to speculate on the changes in the status of bone mineral density after treatment, which can reduce the burden of osteoporotic fractureexpenditure on osteoporosis.
Keyword Osteoporosis Osteopenia Bone fracture Fracture Risk Assessment Tool Bone mineral density Deep learning Convolutional neural network Dual-energy x-ray absorptiometry Precision Medicine Geriatrics
  • Contact
  • Chen, Yueh-Peng
other people also saw