Cone-Beam Computed Tomography as a Prediction Tool for Osteoporosis in Postmenopausal Women (OsteoAI)

Project Objective and Tasks

The objective of the project is to develop a new approach for the early detection of osteoporosis risk in postmenopausal women using cone-beam computed tomography (CBCT) scans. The study will analyse qualitative and morphological changes in the mandibular cortical bone to identify features that may be associated with an increased risk of osteoporosis. Based on the collected data, a comprehensive deep learning prototype model will be developed, with the potential to support automated osteoporosis risk assessment in the future.

An important part of the project is also to investigate the impact of osteoporosis on the bone structures of the temporomandibular joint (TMJ) in postmenopausal women. The structural and morphological characteristics of the mandibular condyle will be assessed, including bone density, volume, surface structure and other relevant parameters, in order to determine how osteoporosis affects these anatomical structures.

To achieve the project objectives, CBCT images and related data will be analysed, relevant morphological bone features associated with osteoporosis risk will be identified, and algorithms and deep learning approaches for their assessment will be developed. The resulting model will be trained and tested to evaluate its ability to identify features indicative of osteoporosis risk. At the same time, the relationship between osteoporosis and changes in TMJ bone structures will be investigated. A total of 250 patients are expected to be included in the study.

Overall, the project aims to combine medical imaging, bone structure analysis and artificial intelligence to develop a potentially more accessible and automated approach for the early identification of osteoporosis risk using CBCT scans obtained in dental care.

Tasks to be Performed by EDI

The contribution of the Institute of Electronics and Computer Science (EDI) to the project focuses on the development of artificial intelligence and image analysis solutions for osteoporosis risk assessment. EDI's tasks cover the full development cycle of the technological solution, from the analysis of morphological features and algorithm development to software implementation, model training and performance evaluation.

As part of the project, EDI will develop a mandibular index model capable of analysing CBCT images and identifying morphological features of the mandibular bone associated with osteoporosis risk.

The tasks to be performed by EDI include:

1. Development of morphological analysis methods and the artificial intelligence model – morphological features of the mandibular bone identifiable in CBCT images will be analysed, and an algorithm for determining mandibular indices will be developed. Based on the obtained data, an appropriate deep neural network architecture will be designed for osteoporosis risk detection.
2. Software development – a software solution will be developed to enable the application of the mandibular index model for CBCT image analysis and osteoporosis risk assessment.
3. Model training and testing – the developed deep learning model will be trained using the data collected within the study and subsequently tested to evaluate its accuracy and suitability for osteoporosis risk assessment.

Overall, EDI's role is to develop a technological solution capable of automatically extracting relevant features for osteoporosis risk assessment from CBCT images and using deep learning methods to provide an automated risk assessment.

The project “RSU Internal and RSU–LSPA External Consolidation” (No. 5.2.1.1.i.0/2/24/I/CFLA/005) is funded under the European Union Recovery and Resilience Facility Plan and the state budget.