Prediction of serosal invasion in gastric cancer: development and validation of multivariate models integrating preoperative clinicopathological features and radiographic findings based on late arterial phase CT images

Table 3 The diagnostic performance of tumor markers, the CT value-related parameters, and texture parameters in the primary cohort

Parameters	Cutoff	Sensitivity	Specificity	AUC	Accuracy	P value
Tumor markers
CA199 (U/mL)	9.47	0.720	0.615	0.681	0.640	0.009*
CA724 (U/mL)	1.48	0.840	0.436	0.652	0.534	0.022*
CA242 (U/mL)	6.30	0.520	0.821	0.662	0.748	0.021*
CT value-related parameters
AP value mean (HU)	96.62	0.680	0.692	0.651	0.689	0.031*
AP value min (HU)	78.00	0.680	0.680	0.658	0.680	0.024*
Δmean A-N (HU)	54.11	0.640	0.705	0.647	0.689	0.035*
Δmean P-A (HU)	−7.46	0.560	0.744	0.650	0.699	0.025*
Δmean D-A (HU)	−25.09	0.680	0.603	0.654	0.622	0.021*
Texture parameters
Mean (HU)	90.74	0.520	0.782	0.648	0.718	0.036*
Max frequency	20.00	0.760	0.680	0.742	0.699	< 0.001*
Minimum (HU)	56.00	0.800	0.539	0.682	0.602	0.005*
5th percentile (HU)	69.00	0.640	0.680	0.662	0.670	0.014*
10th percentile (HU)	70.00	0.560	0.744	0.657	0.699	0.021*
25th percentile (HU)	89.00	0.720	0.603	0.659	0.631	0.021*
50th percentile (HU)	90.00	0.520	0.769	0.645	0.709	0.038*
75th percentile (HU)	101.00	0.520	0.782	0.647	0.718	0.038*
90th percentile (HU)	107.00	0.440	0.846	0.633	0.747	0.064

AUC area under the receiver operating characteristic (ROC) curve, CA carbohydrate antigen, AP arterial phase
*P < 0.05 with ROC curve analysis

ISSN: 1471-2407