Automatic detection of the mental foramen for estimating mandibular cortical width in dental panoramic radiographs

Abstract

Screening tests are vital for detecting diseases, especially at early stages, where efforts can prevent further illness. For example, osteoporosis is a systemic skeletal disease characterized by low bone mass and microarchitectural deterioration of bone tissue, resulting in bone fragility and susceptibility to fracture. Dual-energy x-ray absorptiometry is commonly used to diagnose osteoporosis since it evaluates bone mineral density. It is the most standard method for diagnosing osteoporosis, but it is not immediately available and is commonly used for research due to the high capital cost. Further, dual-energy x-ray absorptiometry is not used for populational-based screening due to its suboptimal ability to predict hip fractures based on measurements. Therefore, it is recommended to adopt a case-finding strategy to identify individuals at risk who benefit from the dual-energy x-ray absorptiometry examination.

Several indices have been developed to estimate bone quality in dental panoramic radiographs to identify individuals at risk of osteoporosis. In particular, the mandibular cortical width index. Studies suggest that dentists can measure the mandibular cortical width to identify individuals at risk and refer them for bone mineral density testing. However, this endeavor is time-consuming and inconsistent due to the bone's unclear borders and the challenge of determining the mental foramen's position, leading to varying measurements between clinicians. Therefore, the dentistry community is investigating how to automate this process effectively and accurately.

In an attempt to address some of these problems, this thesis presents a method to assess the mandibular cortical width index automatically. Four different object detectors were analyzed to determine the mental foramen's position. EfficientDet showed the highest average precision (0.30). Therefore, it was combined with an iterative procedure to estimate mandibular cortical width. The results are promising.