PREDICTION OF PROSTATE BIOPSY OUTCOME BY MULTIPARAMETRIC ULTRASOUND IMAGING
Massimo Mischi,Peiran Chen,Simona Turco,Wim Zwart,Pintong Huang,Hessel Wijkstra
DOI: https://doi.org/10.1016/j.ultrasmedbio.2022.04.037
IF: 3.694
2022-01-01
Ultrasound in Medicine & Biology
Abstract:ObjectivesProstate cancer (PCa) diagnosis still relies on 12-core systematic biopsy (SBx). More recently, pre-biopsy multiparametric MRI has been introduced in the guidelines as a valuable imaging tool for the detection of significant PCa. However, poor reproducibility and specificity, along with its high cost and limited availability, evidence the need for cost-effective, widespread imaging alternatives. 3D contrast-enhanced ultrasound (CEUS) has shown promise for PCa localization by extraction of quantitative perfusion and dispersion features associated with cancer angiogenesis. Tissue stiffness is an additional PCa biomarker that can be quantified by ultrasound shear-wave elastography (SWE). In this study, the diagnostic potential of multiparametric ultrasound imaging, combining 3D CEUS and SWE features, was evaluated for PCa localization by comparison with the corresponding SBx outcome.MaterialsAfter signing an informed consent, 54 biopsy-naïve patients underwent a 3D CEUS recording using a LOGIQ E9 scanner (GE HealthCare, USA) and a multi-plane (probe sweep) 2D SWE scan using an Aixplorer scanner (SuperSonic Imagine, France) at the Second Affiliated Hospital of Zhejiang University (China). All patients received SBx. Multiple 3D perfusion and dispersion feature maps were extracted from the 3D CEUS acquisitions and further complemented with 3D maps of tissue stiffness, reconstructed based on the multi-plane 2D SWE acquisitions. Subsequently, all the 3D maps were subdivided in 12 regions corresponding to the SBx locations. Gradient Boosting (GB) classification was then implemented together with sequential floating forward feature selection for prediction of the biopsy outcome. The classification accuracy was assessed by a k-fold cross-validation procedure.ResultsOf the 54 patients, 20 had SBx-proven significant PCa. Prediction of biopsy outcome using only CEUS features produced an area under the receiver-operating-characteristic curve (AUC) of 0.81, while using the SWE elasticity alone produced an AUC of 0.66. Their combination yielded an improved AUC of 0.85.ConclusionsThe proposed 3D multiparametric ultrasound imaging approach yields accurate PCa classification results, showing promise for cost-effective PCa localization by multiparametric ultrasound imaging. A multicenter trial has recently started to validate and optimize the classification performance in a larger patient cohort.