头颈部腺样囊性癌肺转移结节人工智能影像学研究

Imaging study of artificial intelligence of lung metastatic nodule of adenoid cystic carcinoma in head and neck

目的探讨人工智能(AI)在腺样囊性癌(ACC)肺转移结节定量分析中的应用。方法收集45例ACC患者50个肺转移结节,利用AI软件测量肺转移结节平均直径、最大直径、结节体积,并分别与计算得到的进行配对t检验比较。结果50个肺转移结节均为实性结节。AI测得的结节最大直径、平均直径、体积中位数分别为8.6mm(5.9~11.8mm)、7.2mm(4.5~10.4mm)、81.4mm^(3)(62.9~125.4mm^(3))。AI软件测量体积与人工计算体积在平均直径上的结果相近[(86.9±3.51)mm^(3)与(89.50±5.14)mm^(3),P=0.164)],在最大直径上的结果不同[(103.0±4.60)mm^(3)与(89.50±5.14)mm^(3),P=0.001]。结论AI测量ACC肺转移结节体积较直径能很好地反映结节大小变化,在结节定量分析中有重要临床参考价值,有助于临床决策治疗及随访。

Objective To investigate the quantitative analysis of artificial intelligence(AI)software in metastatic lung nodular size measurement of patients with adenoid cystic carcinoma.Methods 5o lung metastatic nodules of 45 cases of adenoid cystic carcinoma were collected in this study.Artificial intelligence software was used to measure the mean diameter,maximum diameter and nodule volume of pulmonary metastatic nodules and compared with the calculated pairwise t test,Results The median of maximum diameter,mean diameter and median volume of the nodules were 8.6 mm(5.9-11.8 mm),7.2 mm(4.5-10.4 mm)and 1.4mm(62.9-125.4mm^(3))by AI measured,respectively.The results of AI software measurement volume and manual calculation volume are similar to those of average diameter[(86.9±3.51)mm^(3) and(89.50±5.14)mm^(3),P=0.164].The results of maximum diameter are different[(103.0±4.60)mmand(89.50±5.14)mm,P=0.001].ConclusionsVolume of AI measurement is superior to that based on diameter of lung size in quantitative analysis of lung metastatic nodule,which has more important clinic value,and may be helpful in management and follow-up of patients with adenoid cystic carcinoma.