Performance analysis of DMF teeth detection using deep learning: A comparative study with clinical examination as quasi experimental study

Abstract

Introduction: Decayed, missing, and filled teeth (DMF-T) are indicators used to assess the oral health status of an individual or a population. This examination is typically performed manually by dentists or dental therapists. In previous research, researchers have developed a deep learning model as a part of artificial intelligence that can detect DMF-T. Aim of this research was to analyze the comparison of the performance of deep learning with clinical examinations in DMF-T assessment. Methods: Experienced dentists conducted clinical examinations on 50 subjects who met the inclusion criteria. Oral clinical photos of the same patients were taken from various aspects, in total 250 images, and further analyzed using a deep learning model. The results of the clinical examination and deep learning were then statistically analyzed using an unpaired t-test to determine whether there were differences between groups. Results: The unpaired t-test analysis indicated that there was no significant difference between the result of DMF-T examination by dentist and by DL (P>0.05). Unpaired t-test of this research indicated no significant difference (P = 0.161). The unpaired t-test concluded that t Stat < t Critical two-tail, then who was accepted, which stated that there was no significant difference between the results of the DMF-T examination between two groups. Conclusion: The DL model demonstrates good clinical performance in detecting DMF-T.KEYWORDS DMF-T, clinical assessment, deep learning, caries detection