Osteoporosis is a chronic osteometabolic disease characterized by reduced bone mineral density (BMD) and increased risk of fractures. The most up-to-date definitions of osteoporosis introduce the concept of bone quality, which can be interpreted as a generic term describing a set of structural and compositional characteristics of bone responsible for maintaining skeletal resistance to fractures. By definition, fragility fractures are those that occur with low-energy trauma, such as a fall from standing height or less. As the name itself describes, they reflect skeletal fragility to biomechanical stress. Individuals affected by a fragility fracture are at increased risk of suffering other secondary fractures, especially in the first two years after the first fracture occurs. Quantification of bone mass using dual-energy X-ray absorptiometry (DXA) is the “gold standard” test for the diagnosis of osteoporosis. The assessment of bone mass by DXA has been widely validated in clinical trials and is currently the most important tool for the diagnosis of osteoporosis and treatment monitoring. However, its ability to predict the risk of fractures in the presence of osteoporosis is still considered limited. The study presented here trained a ResNet50 convolutional neural network model that was able to characterize the presence of bone fragility and the consequent tendency for vertebral fractures to occur with an accuracy of approximately 78%, specificity of 78%, and sensitivity of 76% in T1-weighted magnetic resonance images of vertebral bodies of patients with osteoporosis and osteopenia.