Applications of NaF-PET and Artificial Intelligence in Osteoporosis

Introduction: 1. Evaluate the utility of <del cite="mailto:Mona-Elisabeth%20Rootwelt-Revheim" datetime="2025-01-11T12:17"> </del>¹⁸F-sodium fluoride (NaF)-PET/CT in assessing osteoporosis.

2. Assess the feasibility of artificial intelligence in diagnosing and managing osteoporosis.

Methods: A thorough literature review was conducted using PubMed and Google Scholar on NaF-PET applications in osteoporosis. The search terms used were "NaF-PET", "osteoporosis", "machine learning" and "artificial intelligence". Special focus was given to scientific articles that discussed the ability of AI to aid in the diagnosis and management of osteoporosis.

Results: Osteoporosis is a degenerative bone disease characterized by loss of bone strength and density predisposing individuals to fractures. The current standard of diagnosing osteoporosis involves dual-energy X-ray absorptiometry (DXA) to assess bone mineral density (BMD) in key areas of degeneration such as the spine and hip. However, DXA scans can only assess bone density without providing information on bone quality such as microarchitectural or material properties, which are key factors in assessing fracture risk. NaF is a radiotracer that accurately assesses bone metabolism through quantifying new bone formation and has been shown to accurately reflect changes in BMD seen in osteoporosis. By quantifying bone metabolic changes, NaF-PET is a sensitive tool in detecting early metabolic changes associated with bone remodeling before BMD changes become apparent.

With the emergence of AI in diagnostic imaging, studies have shown its potential in evaluating osteoporosis-related changes in imaging modalities such as DXA and CT, as well as automating segmentation and predicting fracture risk.AI has been used to predict bone health and progression, classify osteoporosis, and predict outcomes, enabling early intervention and improving prognosis.

Specific to the application of<del cite="mailto:Mona-Elisabeth%20Rootwelt-Revheim" datetime="2025-01-11T12:22"> </del> NaF-PET in osteoporosis, AI can enhance the diagnostic ability through automating segmentation and correlating metabolic findings with clinical risk factors such as body mass index (BMI), age, BMD, and post-menopausal state. By removing the task of manual segmentation in PET analysis, AI enables total body bone metabolic analysis, allowing for more accurate quantification of bone turnover in the entire skeleton rather than limiting analysis to predefined regions of interest. Additionally, AI-driven PET analysis can potentially allow for creation of tools for easy monitoring of treatment response in patients undergoing osteoporosis therapy. AI enables the easy and automatic quantification of metabolic changes over time, allowing clinicians to more efficiently evaluate the effectiveness of different pharmacological interventions.

Limitations, however, lie in the availability of robust, standardized datasets that can train and validate AI models that will perform under a variety of conditions. Different imaging protocols and patient populations can introduce variability that can affect model performance. Further research is needed in training such models and assessing their ability to accurately predict metabolic changes seen in patients with osteoporosis.

Conclusions: AI has shown to have a beneficial role in assessing osteoporotic changes in many imaging modalities. Traditional modalities such as CT and DXA are limited in their ability to only detect structural changes through BMD. NaF-PET, however, can quantify metabolic changes in bone turnover, which allows for early diagnosis of osteoporosis that can be further supplemented by AI. This can allow for global bone metabolic analysis, streamlined monitoring of osteoporosis treatment response, and predictive diagnosis of osteoporosis especially when combined with clinical risk factors such as BMI and age.