Deep learning in the diagnosis for cystic lesions of the jaws: a review of recent progress.

Cystic lesions of the gnathic bones present challenges in differential diagnosis. In recent years, artificial intelligence (AI) represented by deep learning (DL) has rapidly developed and emerged in the field of dental and maxillofacial radiology (DMFR). Dental radiography provides a rich resource for the study of diagnostic analysis methods for cystic lesions of the jaws and has attracted many researchers. The aim of the current study was to investigate the diagnostic performance of DL for cystic lesions of the jaws. Online searches were done on Google Scholar, PubMed, and IEEE Xplore databases, up to September 2023, with subsequent manual screening for confirmation. The initial search yielded 1862 titles, and 44 studies were ultimately included. All studies used DL methods or tools for the identification of a variable number of maxillofacial cysts. The performance of algorithms with different models varies. Although most of the reviewed studies demonstrated that DL methods have better discriminative performance than clinicians, further development is still needed before routine clinical implementation due to several challenges and limitations such as lack of model interpretability, multicentre data validation, etc. Considering the current limitations and challenges, future studies for the differential diagnosis of cystic lesions of the jaws should follow actual clinical diagnostic scenarios to coordinate study design and enhance the impact of AI in the diagnosis of oral and maxillofacial diseases.

Three-dimensional CBCT images registration method for TMJ based on reconstructed condyle and skull base.

OBJECTIVES: A method was introduced for three-dimensional (3D) cone-beamCT (CBCT) images registration of temporomandibular joint (TMJ). This study aimed to provide quantitative and qualitative analysis of TMJ bone changes in two-dimensional (2D) and 3D and to provide the technique for computer-aided diagnosis of temporomandibular joint disorders in the future. METHODS: 10 TMJ samples of six patients were obtained from Peking University Hospital of Stomatology. Four of the six patients imaged bilateral TMJs and the other two patients only imaged unilateral TMJ. Each sample consisted of two images from the same TMJ taken at different times. First, condyle and skull base were segmented semi-automatically for 3D model reconstruction. Then the segmented condyle and skull base were registered separately. Registration process can be divided into two processes of rough registration and fine registration. Rough registration step was achieved by selecting corresponding points manually and initialized fine registration. Condyle and skull base were fine registered by minimizing mean square error of condyle (MSEcondyle) and skull base (MSEskull) respectively. Qualitative assessment of osseous component changes utilized 2D color-fused model and 3D surface-fused model and quantitative analyses the convergence of this method used the mean square error of the model (MSEmodel). Independent repeated experiments were carried out to test the stability of our 3D registration method. RESULTS: Sufficiently alignment was achieved. Osseous abnormality and morphology changes were displayed using fusion model. MSEmodel of condylar registration and skull base registration declined 51.80% and 64.58% compared with that before registration. Quantitative analysis verified the stability of the method. CONCLUSIONS: The proposed method completed 3D TMJ registration for different physiological structure. The result of this method was accurate, reproducible and not relied on the experience of operators.