| Objectives:To compare the diagnostic efficacy of S-Detect and sonographers with different seniority in the diagnosis of benign and malignant thyroid nodules,to evaluate the auxiliary value of S-Detect in the diagnosis of different seniority sonographers,and to explore the clinical application value of different combined diagnostic methods in the diagnosis of benign and malignant thyroid nodules.Methods:1.A total of 180 patients with 206 thyroid nodules,all with Fine-needle aspiration(FNA)for postoperative pathological results,who underwent thyroid preoperative ultrasound examination in our hospital from November 2021 to November 2022 were selected.2.A highly trained sonographer with 9 years of work experience continuously scanned the thyroid and surrounding areas in multiple sections,and the most characteristic transverse section,longitudinal section and color Doppler images of the nodules were collected successively.3.Switch to S-Detect mode,display clear lesions in the transverse and longitudinal sections of the thyroid,and activate the S-Detect function key after freezing to output risk stratification through artificial intelligence(AI)algorithms,making a binary diagnosis of "possibly benign" or "possibly malignant" for thyroid nodules.When the benign and malignant diagnosis results of the transverse and longitudinal sections are consistent,the original diagnosis result is maintained unchanged;when they are inconsistent,to avoid over-diagnosis,it is considered benign.4.The remaining routine ultrasound images of thyroid nodules were analyzed by four sonographers with different seniority(C1,C2,C3,C4 with 1 year,3 years,9years,15 years of working experience in thyroid ultrasound,respectively,and they were not the operators of ultrasound screening).According to the 2020 Chinese version of thyroid imaging reporting and data system(C-TIRADS)classification standard,the characteristics of nodules were scored according to the malignant ultrasound characteristics of nodules.The total score was used for risk stratification,and benign and malignant diagnosis was made(≥ 4B class was considered malignant,and ≤ 4A class was considered benign).5.Adjust the C-TIRADS classification of nodules based on the results of S-Detect and re-evaluate the benign and malignant nature of the nodules.Nodules classified as ≥ 4B are considered malignant,while those classified as ≤ 4A are considered benign.To avoid missed diagnosis and misdiagnosis,two methods are used for joint diagnosis.Method 1: If the S-Detect result suggests malignancy,the C-TIRADS diagnosis is increased by one level(not exceeding 5),while if the S-Detect result suggests benignity,the C-TIRADS diagnosis remains unchanged.Method 2: If the S-Detect result suggests malignancy,the C-TIRADS diagnosis is increased by one level(not exceeding 5),while if the S-Detect result suggests benignity,the C-TIRADS diagnosis is decreased by one level(not below 2).6.The accuracy,sensitivity and specificity of each group were calculated by paired four-table chi-square test.The positive predictive value(PPV)and negative predictive value(NPV)were calculated by generalized score statistical method.The receiver operating characteristic(ROC)curve was drawn with FNA or postoperative pathological results as the gold standard,and the area under the curve(AUC)was calculated to compare the diagnostic efficacy of different groups.At the same time,Kappa coefficient was used to evaluate the consistency between the diagnostic results of different grouping diagnostic methods and postoperative pathological results.Results:1.Among the 206 thyroid nodules,113 cases(113/206),accounting for 54.8%,were confirmed as benign nodules,84 cases(84/206),accounting for 40.8%,were confirmed as malignant nodules,after surgery or FNA pathology.There were 9 cases(9/206)accounting for 4.4%,with inconclusive results from FNA.There were 84 malignant nodules,all of them were papillary thyroid cancer(PTC).Among the 113 cases were benign nodules,7 cases of follicular adenomas(FA),104 cases of multinodular goiters(MNG),1 cases of subacute thyroiditis(SAT),and 1cases of Hürthle cell adenoma(HCA).2.S-Detect has a high sensitivity of 91.7%.S-Detect has a diagnostic accuracy,specificity,sensitivity,PPV and NPV were 78.7%,69.0%,91.7%,68.8% and 91.8%respectively,which were higher than junior sonographer C1(1 year working experience in ultrasound)and sonographer C2(3 year working experience in ultrasound).The accuracy,specificity and PPV of S-Detect were lower than intermediate seniority sonographer C3(9 year working experience in ultrasound)and senior sonographer C4(15 year working experience in ultrasound).The NPV of S-Detect was lower than that of senior sonographer C4 but higher than that of the intermediate seniority sonographer C3.The AUC value of S-Detect was 0.803,which was significantly higher than junior sonographer C1 and junior sonographer C2.It was similar to intermediate seniority sonographer C3,but lower than experienced senior sonographer C4.The Kappa coefficient of S-Detect was 0.582,and the confficiency was general,which higher than that of junior sonographer C1 and sonographer C2,but lower than that of intermediate seniority sonographer C3 and senior sonographer C4.3.By using Method 1 combined diagnosis,the sensitivity of the four sonographers was significantly improved.The accuracy,PPV and NPV of junior sonographer C1(1 year working experience in ultrasound)and C2(3 year working experience in ultrasound)were improved compared with that before combined diagnosis,the accuracy and NPV of intermediate seniority sonographer C3(9 year working experience in ultrasound)were improved,and the NPV of senior sonographer C4(15 year working experience in ultrasound)was improved.The AUC value of junior sonographer C1 and junior sonographer C2 and intermediate seniority sonographer C3 were higher than that before combination,while that of senior sonographer C4 was the same.After combined diagnosis,the Kappa coefficient of junior sonographer C1,sonographer C2 and intermediate seniority sonographer C3 were 0.267,0.596 and 0.631,respectively,and higher than those of the three diagnosis alone.4.By using Method 2,the diagnostic accuracy,specificity,sensitivity,PPV and NPV of junior sonographers C1(1 year working experience in ultrasound)and C2(3year working experience in ultrasound)were improved;the diagnostic accuracy,sensitivity,and NPV of intermediate seniority sonographers C3(9 year working experience in ultrasound)were improved;the diagnostic PPV and NPV of senior sonographer C4(15 year working experience in ultrasound)were improved;The AUC value of junior sonographers C1,junior sonographers C2 and intermediate seniority sonographer C3 were higher than before combined diagnosis,while the AUC value of senior sonographer C4 slightly decreased.After combined diagnosis,the Kappa coefficient of junior sonographer C1 and sonographer C2 and intermediate seniority sonographer C3 were 0.582,0.601 and 0.629,respectively,the confficiency were general,and higher than those of the three diagnosis alone.Conclusions:1.The ease of operation and high sensitivity of S-Detect can provide auxiliary diagnosis for conventional ultrasounds.2.S-Detect can achieve the diagnostic level of an intermediate seniority sonographer with 9 years of experience.3.The combination of S-Detect and C-TIRADS classification can improve the diagnostic efficiency of junior sonographers,while the auxiliary diagnostic value for experienced senior sonographers is limited.4.The combination of the two methods can improve the diagnostic efficiency of intermediate seniority and junior sonographers.The first method can significantly improve the diagnostic sensitivity of sonographers,while the second method is better in improving the diagnostic efficiency of junior sonographers. |
PDF Full Text Request