Iodine Maps from Dual-Energy CT to Predict Extrathyroidal Extension and Recurrence in Papillary Thyroid Cancer Based on a Radiomics Approach

Article Figures & Data

Download figure
Open in new tab
Download powerpoint
FIG 1.
Our study flow chart and radiomics workflow.
Download figure
Open in new tab
Download powerpoint
FIG 2.
Kaplan-Meier cumulative event curve for PTC recurrence stratified by the nomogram-predicted ETE risk classification (log-rank statistic, P < . 001).

Table 1:

Results of multivariate logistic analysis for constructed clinical and radiomics models

Models	β	Odds Ratio (95% CI)	P Value	VIF
Clinical model
Size	0.04	1.03 (0.98–1.09)	.009	1.61
Capsule contact	1.53	3.55 (2.32–5.44)	<.001	1.68
Radiomics model 1
Size	0.04	1.04 (0.98–1.10)	.041	1.57
Capsule contact	1.29	3.61 (2.34–5.59)	<.001	1.56
Noncontrast rad-score_RF	–0.51	0.60 (0.44–0.83)	.002	1.00
Radiomics model 2
Size	0.04	1.04 (0.98–1.09)	.045	1.57
Capsule contact	1.28	3.58 (2.33–5.50)	.001	1.56
Noncontrast rad-score_LASSO	–0.43	0.65 (0.41–1.04)	.007	1.01
Radiomics model 3
Capsule contact	1.26	3.51 (2.41–5.13)	<.001	1.62
Mixed rad-score_RF	2.86	17.44 (1.71–177.60)	<.001	1.30
Radiomics model 4
Size	0.02	1.02 (0.97–1.08)	.036	1.83
Capsule contact	1.24	3.45 (2.24–5.33)	<.001	1.47
Mixed rad-score_LASSO	1.49	14.44 (1.05–120.93)	<.001	1.51
Radiomics model 5
Size	0.01	1.00 (0.95–1.06)	.034	1.78
Capsule contact	1.11	3.03 (1.95–4.71)	<.001	1.69
Iodine rad-score_RF	3.63	37.64 (6.76–209.59)	<.001	1.34
Radiomics model 6
Size	0.01	0.99 (0.94–1.05)	.048	1.74
Capsule contact	1.09	2.98 (1.92–4.64)	<.001	1.62
Iodine rad-score_LASSO	2.86	35.32 (2.29–289.19)	<.001	1.37

Table 2:

Prediction efficiency comparisons of 6 radiomics models and the clinical model

Radiomics vs Clinical Models (AUC)	Training Set (317)	P Value	Validation Set (135)	P Value
Radiomics model 1 vs clinical model	0.73 vs 0.72	.920	0.73 vs 0.72	.838
Radiomics model 2 vs clinical model	0.73 vs 0.72	.877	0.73 vs 0.72	.777
Radiomics model 3 vs clinical model	0.76 vs 0.72	.022	0.76 vs 0.72	.132
Radiomics model 4 vs clinical model	0.75 vs 0.72	.064	0.74 vs 0.72	.179
Radiomics model 5 vs clinical model	0.78 vs 0.72	<.001	0.84 vs 0.72	<.001
Radiomics model 6 vs clinical model	0.77 vs 0.72	<.001	0.80 vs 0.72	.032

Table 3:

Univariable and multivariable Cox regression analysis for risk factors associated with recurrence