Sensitivitas dan Spesifisitas Klasifikasi Maloklusi Ditinjau dari Radiograf Sefalometri Menggunakan Software Yolo V4

Abstract

Sensitivity and Specificity of Malocclusion Classification Reviewed from Cephalometric Radiographs Using YOLO V4 Software xiii + 40 pages Lateral cephalometric radiography plays a crucial role in the analysis of skeletal relationships; however, manual measurements are prone to operator subjectivity and are relatively time-consuming. Advances in artificial intelligence, particularly You Only Look Once version 4 (YOLO v4), have the potential to improve the accuracy and efficiency of cephalometric analysis. This study aimed to determine the mean values of SNA, SNB, and ANB angles in Class I, II, and III malocclusions measured using YOLO v4 software and manual cephalometric analysis, as well as to evaluate the sensitivity and specificity of YOLO v4. This analytical cross-sectional study involved 330 lateral cephalometric radiographs of patients aged 20–30 years with malocclusion. Angular measurements were performed using manual cephalometric analysis and YOLO v4 software. The results showed no significant differences in the mean SNA, SNB, and ANB values obtained using YOLO v4 compared with manual cephalometric analysis across all malocclusion classes (p > 0.05). The YOLO v4 software demonstrated a mean specificity of 0.7688 and an accuracy of 100%. It can be concluded that YOLO v4 exhibits good sensitivity and specificity, with measurement outcomes comparable to manual cephalometric analysis. Therefore, YOLO v4 has the potential to be used as an efficient and objective diagnostic aid for malocclusion assessment in orthodontic practice