Can MRI predict subsequent pseudarthrosis resulting from osteoporotic thoracolumbar vertebral fractures?