PREDICTION OF MALIGNANCY WITH RGB PROFILING OF PANCREATIC MASS- ELASTOGRAPHIES AND CONTRAST-ENHANCED IMAGES BY ENDOSCOPIC ULTRASOUND

Aims Combining elastography and contrast-enhanced endoscopic ultrasound (CE-EUS) is useful in the semi-quantitative evaluation of pancreatic tumors. We aimed to quantify the CE images and elastographies obtained by EUS to develop a prediction model for malignancy of solid pancreatic lesions.

Methods Quantitative analysis of elastographies and CE images was performed by RGB (red-green-blue) profiling using Java-based processing software (ImageJ, NIH). The exact amount of red (soft-), green (intermediate-), blue (hard-tissue), and enhanced areas (viable tumor) at peak enhancement, was measured and expressed in pixels. The intensity ratio for each color was defined as a relation between the absolute value for this color and the intensity of the sum of all three colors (R,G, and B Int.- Ratio). The proportion between enhanced (HyperE) and non-enhanced (HypoE) areas was calculated (HH-Ratio). The final diagnosis was established either by histopathology or radiological findings combined with tumor markers and clinical follow-up.

Results Between 01/2014-08/2021, we identified 88 solid pancreatic tumors examined by elastography and CE-EUS during the same procedure: 63.6% adenocarcinomas, 8% metastasis, 3.4% neuroendocrine tumors, 4.5% other malignant tumors, 18 (20.5%) benign masses. For parameters with the best performance to differentiate between malignant and benign pancreatic tumors, areas under the receiver-operating characteristic curves with the best cut-off values were calculated ([Table 1]). The accuracy of these criteria to predict malignancy was: 3 to 5 criteria (n=32)-96.9%, 1 or 2 criteria (n=37)-75.7%, 0 criteria (n=19)-50%.

Conclusions Quantitative analysis of elastographies and contrast-enhanced images of pancreatic tumors obtained in EUS has the potential to predict malignancy with high accuracy.

Table 1

Variable	Correlation (p)	Criterion (AUC; 95% CI)	Sensitivity (%)	Specificity (%)
B Int.-Ratio	r=0,361 (0,0007)	>230 (0,750; 0,645-0,838)	56,2	89,5
B/G Int.-Ratio	r=0,340 (0,001)	>1,9 (0,735; 0,628-0,825)	62,1	79,0
G Int.-Ratio	r=-0,307 (0,004)	<97,4 (0,713; 0,605-0,806)	34,9	100
HyperE	r=-0,317 (0,003)	<11,9 (0,693; 0,583-0,788)	60,7	75,9
HH-Ratio	r=-0,334 (0,001)	<16,4 (0,703; 0,594-0,797)	55,4	86,2

Publication History

Article published online:
14 April 2022