Computer-Aided Diagnosis For Liver Trauma With Active Bleeding By Contrast-enhanced Ultrasound: An Experimental Study

Objective:1. To investigate the value of Contrast-enhanced Microvascular Imaging (CEMVI) in the diagnosis of liver traumatic active bleeding; 2. To evaluate the value of Three-dimensional Contrast-Enhanced Sonography (3D-CEUS) in the diagnosis of liver trauma with active bleeding.3. To evaluate the value of Computer-Aided Dignosis Contrast-enhanced Ultrasound (CAD-CEUS) in the grading of liver traumatic active bleeding.Materials and Methods:1. Experiment one:21 liver traumatic active bleeding were made and in 8 pigs. The grading was made according to the AAST standards. CEUS and CEMVI were performed immediately to observe the sonographic characteristics of the active bleeding.2. Experiment two:31 liver traumatic lesions with 21 active bleeding sites were made in 8 pigs. The grading was made according to the AAST standards. CEUS and 3D-CEUS were performed immediately to observe the sonographic characteristics of the traumatic lesion and active bleeding, the sensitivity and specificity were compared between the two methods.3. Experiment three:14 traumatic lesions with 11 active bleedings were made in 14 pigs. CEUS and CAD-CEUS were performed to classify the liver trauma and active bleeding according to the CEUS and CAD-CEUS grading criteria. Difference was analyzed for the hemorrhagic shock occurring time of pigs under different CAD-CEUS grading.Results:1. Experiment one:CEUS showed 17(60.7%) bleeding sites inⅠ～Ⅳtrauma, with the feature of "fountain-like gush" or "nodular enhanced regions" or "seeping-out intermittent leaking". CEMVI showed 24(85.7%) bleeding sites inⅠ～Ⅳtrauma, with more remarkable moving track features compared with CEUS. CEMVI performed higher sensitivity for detecting active bleeding than CEUS (P<0.05). There was good consistency between two investigators.2. Experiment two: 31 liver traumatic lesions with 21 active bleeding, CEUS and 3D-CEUS showed all traumatic lesions as nonperfusion area in all phases.3D-CEUS could additionally show the volume data of lesions. CEUS showed 16 bleeding sites and misjudged 4 normal tissues.3D-CEUS showed 18 bleeding sites without misjudging normal tissues. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy for CEUS were 76.2%,60%,80%,54.5%,70.9%, and for 3D-CEUS,85.7%,100%,100%,76.9%,90.3%, respectively.3D-CEUS performed better specificity and negative predictive value (P<0.05).6(28.6%) broken vessel section were seen by 3D-CEUS. There was good consistency between two investigators.3. Experiment three:the hemorrhagic shock was occurred in 9 pigs 1.8±0.9 hours after trauma creation and died ultimately. CEUS showed 9 bleeding sites. CAD-CEUS showed and classified 10 in 14 bleeding sites effectively. The hemorrhagic shock occurring times of pigs were different under CAD-CEUS grading. There was good consistency between two investigators.Conclusions:1. CEMVI could be used to show liver traumatic active bleeding as the characters of unlimited movement mode, especially for "seeping-out intermittent leaking" and performed higher sensitivity compared with CEUS, which was helpful to the diagnosis of liver traumatic active bleeding.2.3D-CEUS offered better specificity and negative predictive value and had the ability of volume measurement and showing spatial relationships and surrounding structures in detecting bleeding site, which might be useful in the diagnosis of blunt liver trauma with active bleeding.3. CAD-CEUS could classify the active bleeding in liver trauma effectively, which was benefit for the evaluation of severity of active bleeding and for the selection of therapeutic ways.