Objective To explore the effect of the increase of the number of cases in the database and the difference of the matching number on the results of automatic delineation of lung cancer endangering organs based on MIM software.Methods five groups of lung cancer databases were established by using MIM software(the number of inbound cases were 20,40,60,80 and 100 respectively).The images of 10 patients with lung cancer outside the database were randomly select-ed,and a doctor with rich clinical experience manually sketched the dangerous organs(bilateral thyroid,bilateral lungs,heart,liver and spinal cord),which were defined as reference sketching,and the images of 10 patients were automatically sketched the dangerous organs,and the matching numbers of sketching were selected as 3 and 9 respectively.Through quan-titative evaluation of sketching time,Hausdorff distance(HD),mean minimum distance(MDA),dice similarity coefficient(DSC)and Jaccard coefficient(JSC),the automatic sketching results and reference sketching were analyzed by one-way ANOVA,so as to explore the influence of the number of cases and matches in different databases on the automatic sketching results.Results when the matching number of sketching was 3 or 9,the average automatic sketching time in the five groups was less than that of manual sketching,but there was no difference among the groups;The increase of the number of sketching matches will increase the time of automatic sketching of oar,but it is still lower than the time required for manual sketching of oar.There was no significant difference between the automatic sketching results of the five groups(P＞0.05).For spinal cord,when the matching number is 9,the MDA and JSC results of 100 cases of database automatic drawing re-sults are better than those of 40 cases and 60 cases.But for other organs,there was no significant difference between the au-tomatic sketching results of 3 and 9 different matching numbers in 5 different databases.For thyroid,the results of MDA,DSC and JSC are poor,and the automatic delineation of thyroid can not meet the clinical requirements.Conclusion the application of MIM software in automatic outlining of oar in radiotherapy for lung cancer can achieve better outlining effect and improve work efficiency.The increase of the number of cases included in the database and the number of sketching matches will not improve the sketching effect of oar.At the same time,the increase of the number of cases included in the database will not cause the increase of the time of automatic sketching oar,but the increase of the number of sketching mat-ches will cause the increase of the time of automatic sketching oar.Therefore,considering the time cost,it is recommended to use 20 cases as the number of cases in the clinical database and 3 cases as the template for the sketching of lung cancer.
Lung cancerRadiotherapyMIM maestroOrgans at riskAutomatic delineation
NETL
NSTL
维普
万方数据
