TY - JOUR
T1 - Focal liver lesions
T2 - Computer-aided diagnosis by using contrast-enhanced us cine recordings
AU - Ta, Casey N.
AU - Kono, Yuko
AU - Eghtedari, Mohammad
AU - Oh, Young Taik
AU - Robbin, Michelle L.
AU - Barr, Richard G.
AU - Kummel, Andrew C.
AU - Mattrey, Robert F.
N1 - Funding Information:
This study received support from the National Cancer Institute (F31CA177199, R25CA153915); the Center for Cross Training Translation Cancer Researchers in Nanotechnology (R25 CA153915); the Cancer Prevention Research Institute of Texas (CPRIT-RR150010); and Bracco Diagnostics, which permitted the use of the contrast-enhanced US cine clips originally acquired in their sponsored clinical trial.
Funding Information:
from all subjects enrolled at three independent sites (University of California San Diego, University of Alabama at Birmingham, and Southwoods Imaging [Youngstown, Ohio]) that were part of a Bracco Diagnostics (Princeton, NJ) multicenter trial (https://clinicialtri-als.gov #NCT00788697). The primary study at all three sites had obtained institutional review board approval and written informed consent and admitted any subject with an FLL visible on the baseline US study. It was intended to evaluate the use of SonoVue (Bracco Imaging, Milan, Italy) in FLL characterization. Our retrospective review had institutional review board approval with waiver of informed consent for further analysis of de-identified cine clips, US images, and final diagnoses as benign, malignant, or indeterminate, without subject demographic or clinical data. Sponsorship from Bracco Diagnostics for the primary clinical trial at each site included the provision of financial support and contrast agents to the principle investigators. R.G.B. is on the advisory panel for Bracco. The entire data set from all three sites was provided with Bracco’s consent; however, Bracco neither had control over the data submitted for publication nor reviewed the manuscript prior to submission. C.N.T., A.C.K., and R.F.M. had full control of data and materials submitted for publication. Twenty-two of the cine clips included in this study were also included in our published study on motion correction algorithms (16).
Publisher Copyright:
© RSNA, 2017.
PY - 2018/3
Y1 - 2018/3
N2 - Purpose: To assess the performance of computer-aided diagnosis (CAD) systems and to determine the dominant ultrasonographic (US) features when classifying benign versus malignant focal liver lesions (FLLs) by using contrast material-enhanced US cine clips. Materials and One hundred six US data sets in all subjects enrolled by three centers from a multicenter trial that included 54 malignant, 51 Methods: benign, and one indeterminate FLL were retrospectively analyzed. The 105 benign or malignant lesions were confirmed at histologic examination, contrast-enhanced computed tomography (CT), dynamic contrast-enhanced magnetic resonance (MR) imaging, and/or 6 or more months of clinical follow-up. Data sets included 3-minute cine clips that were automatically corrected for in-plane motion and automatically filtered out frames acquired off plane. B-mode and contrast-specific features were automatically extracted on a pixel-by-pixel basis and analyzed by using an artificial neural network (ANN) and a support vector machine (SVM). Areas under the receiver operating characteristic curve (AUCs) for CAD were compared with those for one experienced and one inexperienced blinded reader. A third observer graded cine quality to assess its effects on CAD performance. Results: CAD, the inexperienced observer, and the experienced observer were able to analyze 95, 100, and 102 cine clips, respectively. The AUCs for the SVM, ANN, and experienced and inexperienced observers were 0.883 (95% confidence interval [CI]: 0.793, 0.940), 0.829 (95% CI: 0.724, 0.901), 0.843 (95% CI: 0.756, 0.903), and 0.702 (95% CI: 0.586, 0.782), respectively; only the difference between SVM and the inexperienced observer was statistically significant. Accuracy improved from 71.3% (67 of 94; 95% CI: 60.6%, 79.8%) to 87.7% (57 of 65; 95% CI: 78.5%, 93.8%) and from 80.9% (76 of 94; 95% CI: 72.3%, 88.3%) to 90.3% (65 of 72; 95% CI: 80.6%, 95.8%) when CAD was in agreement with the inexperienced reader and when it was in agreement with the experienced reader, respectively. B-mode heterogeneity and contrast material washout were the most discriminating features selected by CAD for all iterations. CAD selected time-based time-intensity curve (TIC) features 99.0% (207 of 209) of the time to classify FLLs, versus 1.0% (two of 209) of the time for intensity-based features. None of the 15 video-quality criteria had a statistically significant effect on CAD accuracy-all P values were greater than the Holm-Sidak a-level correction for multiple comparisons. Conclusion: CAD systems classified benign and malignant FLLs with an accuracy similar to that of an expert reader. CAD improved the accuracy of both readers. Time-based features of TIC were more discriminating than intensity-based features.
AB - Purpose: To assess the performance of computer-aided diagnosis (CAD) systems and to determine the dominant ultrasonographic (US) features when classifying benign versus malignant focal liver lesions (FLLs) by using contrast material-enhanced US cine clips. Materials and One hundred six US data sets in all subjects enrolled by three centers from a multicenter trial that included 54 malignant, 51 Methods: benign, and one indeterminate FLL were retrospectively analyzed. The 105 benign or malignant lesions were confirmed at histologic examination, contrast-enhanced computed tomography (CT), dynamic contrast-enhanced magnetic resonance (MR) imaging, and/or 6 or more months of clinical follow-up. Data sets included 3-minute cine clips that were automatically corrected for in-plane motion and automatically filtered out frames acquired off plane. B-mode and contrast-specific features were automatically extracted on a pixel-by-pixel basis and analyzed by using an artificial neural network (ANN) and a support vector machine (SVM). Areas under the receiver operating characteristic curve (AUCs) for CAD were compared with those for one experienced and one inexperienced blinded reader. A third observer graded cine quality to assess its effects on CAD performance. Results: CAD, the inexperienced observer, and the experienced observer were able to analyze 95, 100, and 102 cine clips, respectively. The AUCs for the SVM, ANN, and experienced and inexperienced observers were 0.883 (95% confidence interval [CI]: 0.793, 0.940), 0.829 (95% CI: 0.724, 0.901), 0.843 (95% CI: 0.756, 0.903), and 0.702 (95% CI: 0.586, 0.782), respectively; only the difference between SVM and the inexperienced observer was statistically significant. Accuracy improved from 71.3% (67 of 94; 95% CI: 60.6%, 79.8%) to 87.7% (57 of 65; 95% CI: 78.5%, 93.8%) and from 80.9% (76 of 94; 95% CI: 72.3%, 88.3%) to 90.3% (65 of 72; 95% CI: 80.6%, 95.8%) when CAD was in agreement with the inexperienced reader and when it was in agreement with the experienced reader, respectively. B-mode heterogeneity and contrast material washout were the most discriminating features selected by CAD for all iterations. CAD selected time-based time-intensity curve (TIC) features 99.0% (207 of 209) of the time to classify FLLs, versus 1.0% (two of 209) of the time for intensity-based features. None of the 15 video-quality criteria had a statistically significant effect on CAD accuracy-all P values were greater than the Holm-Sidak a-level correction for multiple comparisons. Conclusion: CAD systems classified benign and malignant FLLs with an accuracy similar to that of an expert reader. CAD improved the accuracy of both readers. Time-based features of TIC were more discriminating than intensity-based features.
UR - http://www.scopus.com/inward/record.url?scp=85042448540&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85042448540&partnerID=8YFLogxK
U2 - 10.1148/radiol.2017170365
DO - 10.1148/radiol.2017170365
M3 - Article
C2 - 29072980
AN - SCOPUS:85042448540
SN - 0033-8419
VL - 286
SP - 1062
EP - 1071
JO - RADIOLOGY
JF - RADIOLOGY
IS - 3
ER -