AI can predict the progression of osteoporosis fractures in the spine
Mar 04, 2025
Professor Park Sung-bae of neurosurgery at Seoul Boramae Hospital, operated by Seoul National University Hospital, announced the results of an AI model that predicted the progression of vertebral osteoporosis fractures using spinal MRI with the KAIST research team.
Osteopathic vertebral compression fracture (OVCF) is a disease that risks further deterioration of the vertebral structure over time, and early prediction is essential. This is because the patient's additional spinal damage can lead to chronic back pain, nerve damage, and spinal deformation, which can seriously deteriorate the patient's quality of life.
However, it is difficult to predict further spinal injury progression early with conventional clinical evaluation alone. X-ray or CT scans alone are less accurate, and there is a limit to relying on the experience of medical staff.
In response, Professor Park Sung-bae's research team supplemented these limitations with the development of AI and MRI-based prediction models.
We retrospectively analyzed the risk of further spinal fracture progression (VC) based on MRI imaging and clinical data in 245 patients with osteoporosis spinal compression fractures (OVCF) between January 2020 and December 2023. Among them, a deep AI-based learning model was developed using data from 200 people, and the predictive performance was verified in a test group of 45 people.
Evaluation of the performance of AI models shows that ViT-PMC-LoRA models have the highest prediction accuracy compared to other conventional models (AUC: 0.8656), and the prediction success rate is further improved, especially when augmented prediction techniques are introduced.
Through this technology, the possibility of early screening of patients at high risk of spinal fracture progression and making appropriate treatment plans has increased.
Professor Park Sung-bae "This study has increased the possibility of screening patients at high risk of spinal fracture at an early stage" and added "Hopefully, the use of this model can increase the diagnostic accuracy of medical staff and provide practical help in establishing treatment plans."
The study was recently published in the journal 『Scientific Reports』.
Osteopathic vertebral compression fracture (OVCF) is a disease that risks further deterioration of the vertebral structure over time, and early prediction is essential. This is because the patient's additional spinal damage can lead to chronic back pain, nerve damage, and spinal deformation, which can seriously deteriorate the patient's quality of life.
However, it is difficult to predict further spinal injury progression early with conventional clinical evaluation alone. X-ray or CT scans alone are less accurate, and there is a limit to relying on the experience of medical staff.
In response, Professor Park Sung-bae's research team supplemented these limitations with the development of AI and MRI-based prediction models.
We retrospectively analyzed the risk of further spinal fracture progression (VC) based on MRI imaging and clinical data in 245 patients with osteoporosis spinal compression fractures (OVCF) between January 2020 and December 2023. Among them, a deep AI-based learning model was developed using data from 200 people, and the predictive performance was verified in a test group of 45 people.
Evaluation of the performance of AI models shows that ViT-PMC-LoRA models have the highest prediction accuracy compared to other conventional models (AUC: 0.8656), and the prediction success rate is further improved, especially when augmented prediction techniques are introduced.
Through this technology, the possibility of early screening of patients at high risk of spinal fracture progression and making appropriate treatment plans has increased.
Professor Park Sung-bae "This study has increased the possibility of screening patients at high risk of spinal fracture at an early stage" and added "Hopefully, the use of this model can increase the diagnostic accuracy of medical staff and provide practical help in establishing treatment plans."
The study was recently published in the journal 『Scientific Reports』.
|
This article was translated by Naver AI translator.
