Discriminant Function Analysis For Parametrial Infiltration In Cervical Carcinoma With CT Perfusion Imaging

Objective To study the efficacy of discriminant function analysis for parametrial infiltration in cervical carcinoma with CT perfusion imaging and to improve the diagnostic accuracy and specificity of parametrial infiltration in cervical carcinoma.Methods64-slice spiral CT perfusion imaging was performed in42cervical carcinoma patients before surgery.The cases were classified into two groups (the non-parametrial infiltration group and the parametrial infiltration group) according to pathological date. The time-density curve (TDC) and blood flow (BF),blood volume (BV), time to start (TTS), time to peak (TTP), permeability and patlak blood volume (PBV) of all regions of interest were measured and statistically analyzed. The discriminant functions were calculated. The parametrial infiltration was determined by discriminant function and CT morphology.The results of two methods were compared with the pathological results.Results The mean BF、BV、TTS、TTP、permeability and pBV were:(63.06±16.05) ml·100ml-1·min-1.(89.13±16.66)ml/L、(38.31±13.87)0.1s、(143.09±19.40)0.1s、(56.36±24.06)0.5ml·100ml-1·min-1and (47.00±24.13)ml/L in the non-parametrial infiltration group. The mean BF、BV、TTS、TTP、permeability and pBV were:(84.67±31.65)ml·100ml-1·min-1,(111.36±36.24)ml/L,(26.62±9.94)0.1s,(123.77±13.20)0.1s,(50.60±14.37)0.5ml·100ml-1·min-1,(102.28±62.77)ml/L in the parametrial infiltration group. The mean BF、BV、TTS、TTP、PBV between the non-parametrial infiltration group and the parametrial infiltration group were statistically significant (P<0.05). The mean permeability was not statistically significant (P>0.05).The discriminant function were obtained as follow:F1(the non-parametrial infiltration group)=-45.34848+0.58933×TTP+0.08499×PBV+0.06200×TTS,F2(the parametrial infiltration group)=-39.24907+0.54863×TTP+0.10448×PBV-0.00341×TTS.Based on the discriminant function,the diagnostic accuracy and specificity of parametrical infiltration in cervical carcinoma were89.5%and90.9%.Based on the CT morphology, the accuracy and specificity were68.0%and63.6%.Conclusion The discriminant function with CT perfusion imaging may improve the diagnostic accuracy and specificity of parametrial infiltration in cervical carcinoma.