Research On Artificial Intelligence Diagnosis System Of Dental Diseases Based On Panoramic Radiograph

Research on Artificial Intelligence Diagnosis System of Dental Diseases Based on Panoramic RadiographObjective Artificial intelligence(AI)has been introduced to interpret the clinical panoramic radiographs of adults.The purpose of the study was to develop a novel deep learning-based AI framework to diagnose multiple dental diseases on panoramic radiographs,and to initially evaluate the preliminary diagnostic performance of the framework in clinical applications.Methods Part I:Three dentists with more than 12 years of clinical experience manually annotated 1996 panoramic radiographs with the following contents:normal teeth,missing teeth,caries,residual roots,full crowns,and impacted teeth.The annotated data sets were randomly divided into training,validation and test sets after the other 2 experts confirmed the uniformity.These data were used to train a new framework based on the BDU-net and nnU-net.In the test set,the sensitivity and specificity of the framework were calculated for tooth loss;sensitivity,specificity,and mean Dice for caries,residual roots,full crowns,and impacted teeth.Agreement between the framework and the experts was determined by the Cohen’s Kappa test.Part Ⅱ:Diagnostic evaluation was performed on a separate dataset including 282 panoramic radiographs.Sensitivity,specificity,Youden’s index,the area under the curve(AUC),and diagnostic time were calculated.Dentists with 3 different levels of seniority(H:high,M:medium,L:low)diagnosed the test panoramic radiographs independently.Mann-Whitney U test and Delong test were conducted for statistical analysis for AUC and the mean diagnostic time,respectively,with a test level of α=0.05.Results Part Ⅰ:On the test set,the sensitivity and specificity of the framework for diagnosing 5 diseases were 0.863,0.983(missing teeth),0.718,0.997(residual roots),0.942,0.986(impacted teeth),0.821,0.989(caries)and 0.835,0.991(full crowns),respectively;and the mean Dice for residual roots,impacted teeth,caries and full crowns were 0.862,0.943,0.841 and 0.926,respectively.Cohen’s Kappa coefficient was greater than 0.80(p<0.001),which means that the framework had a high agreement with the experts.Part Ⅱ:Sensitivity,specificity,and Youden’s index of the framework for diagnosing 5 diseases were 0.964,0.996,0.960(impacted teeth),0.953,0.998,0.951(full crowns),0.871,0.999,0.870(residual roots),0.885,0.994,0.879(missing teeth),and 0.554,0.990,0.544(caries),respectively.AUC of the framework for the diseases were 0.980(95%CI:0.976-0.983,impacted teeth),0.975(95%CI:0.972-0.978,full crowns),and 0.935(95%CI:0.929-0.940,residual roots),0.939(95%CI:0.934-0.944,missing teeth),and 0.772(95%CI:0.764-0.781,caries),respectively.AUC of the AI framework was comparable to that of all dentists in diagnosing residual roots(p>0.05),and its AUC values were similar to(p>0.05)or better than(p<0.05)that of M-level dentists for diagnosing 5 diseases.But AUC of the framework was statistically lower than some of H-level dentists for diagnosing impacted teeth,missing teeth,and caries(p<0.05).The mean diagnostic time of the framework was significantly lower than that of all dentists(p<0.001).Conclusions The new AI framework based on BDU-net and nnU-net demonstrated was effective in segmenting impacted teeth,full crowns,residual roots,and caries on panoramic radiographs,with high efficiency and specificity in diagnosing impacted teeth,missing teeth,full crowns,residual roots,and caries with high efficiency.Its diagnostic performance was similar to or even better than the dentists with 3-10 years of experiences.The AI framework has potential to be applied for diagnosing multiple dental diseases on panoramic radiographs,but the accuracy of diagnosing caries should be improved.