Recursive hierarchic segmentation analysis of bone mineral density changes on digital panoramic images

Area of Science:

• Oral radiology
• Biomedical imaging
• Osteoporosis research

Background:

• Osteoporosis significantly impacts bone health, particularly in postmenopausal women.
• Accurate diagnosis of osteopenia and osteoporosis is crucial for timely intervention.
• Digital panoramic imaging (DPI) offers a potential non-invasive diagnostic tool.

Purpose of the Study:

• To evaluate histogram analysis and mathematical modeling of DPIs for discriminating bone density.
• To assess the efficacy of recursive hierarchic segmentation (RHSEG) in processing DPIs for bone analysis.
• To differentiate between normal, osteopenic, and osteoporotic cancellous bone using image analysis techniques.

Main Methods:

• Analysis of 47 DPIs from postmenopausal women, categorized by dual-energy x-ray absorptiometry.
• Application of RHSEG to mandibular trabecular regions (angle and canine/premolar).
• Histogram analysis, relative intensity function generation, and generalized linear mixed model analysis.

Main Results:

• Histogram analysis successfully discriminated between normal, osteopenic, and osteoporotic bone groups.
• Receiver operating characteristic analysis showed high accuracy (0.78 for osteoporosis, 0.74 for osteopenia) in the canine/premolar region.
• Cubic and quartic models were required for osteoporosis and osteopenia discrimination, respectively, with neither model alone classifying all groups.

Conclusions:

• Histogram analysis and mathematical modeling of RHSEG-processed DPIs can effectively differentiate patients with varying bone densities.
• This imaging analysis approach shows promise for non-invasive diagnosis of osteopenia and osteoporosis.
• Further refinement of mathematical models may enhance diagnostic accuracy across all bone density categories.