[SIST Distinguished Seminar] Prostate cancer diagnostics by multiparametric ultrasound: from microvascular characterization to AI enhancement

ON2024-05-20TAG: ShanghaiTech UniversityCATEGORY: Lecture

Topic: Prostate cancer diagnostics by multiparametric ultrasound: from microvascular characterization to AI enhancement

Speaker: Professor Massimo Mischi, Eindhoven University of Technology (TU/e)

Date and time: May 22, 10:00

Venue: Room A200, #1 Building of SIST

Host: Xu Lin


Abstract:

Prostate cancer (PCa) diagnosis still relies on 12-core systematic biopsy. Recently, pre-biopsy multiparametric (mp) MRI has been introduced in the European guidelines as a valuable imaging tool for PCa detection. However, poor reproducibility and specificity, along with high cost and limited availability, evidence the need for cost-effective, widespread imaging alternatives. Leveraging the association between cancer growth and angiogenesis, dynamic contrast-enhanced ultrasound (DCE-US) has shown promise for PCa localization. We have developed a novel DCE-US imaging solution for the localization of PCa angiogenesis by pharmacokinetic modeling of the convective-dispersion process regulating the contrast transport through tissue. Several estimators of dispersion have been developed over the years by temporal and spatiotemporal analysis of the measured time intensity curves following the intravenous injection of an ultrasound contrast bolus. Targeting biopsies based on quantitative dispersion features associated with PCa angiogenesis, we have achieved PCa detection rates comparable to mpMRI. Tissue stiffness is an additional PCa biomarker that can be quantified by ultrasound shear-wave elastography. We have demonstrated that complementing our dispersion analysis with additional features characterizing tissue structure and mechanical behavior in a machine learning framework can further boost the diagnostic accuracy. Additional advances consisted in the extension of our multiparametric ultrasound (mpUS) analysis from 2D to 3D, facilitating the clinical workflow and improving the accuracy of the estimates based on measured boundary conditions, yet at the cost of a compromise in terms of spatio-temporal resolution. Our latest developments have recently been implemented by a startup company that has completed a large multicenter trial comparing quantitative mpUS, enhanced by deep-learning classification, with the histopathological ground truth through elastic registration. The obtained voxel level classification performance based on over 300 patients evidenced an area under the receiver operator characteristic curve approaching 90%. An additional multicenter study is currently ongoing that aims at demonstrating the non-inferiority of mpUS over mpMRI for targeting prostate biopsies. This seminar will present all the developments that we made in the past years for the localization of PCa by DCE-US and mpUS maging, along with the latest updates on the ongoing clinical trials.


Biography:

Massimo Mischi received the MSc degree in electronic engineering (1999) from La Sapienza University of Rome (Italy) and the PhD degree (2004) from the Eindhoven University of Technology (TU/e, The Netherlands). In 2007, he was an Assistant Professor with the Electrical Engineering Department, TU/e. In 2011, he was an Associate Professor at TU/e and founded the Biomedical Diagnostics Research Laboratory, with focus on model-based quantitative analysis of biosignals with applications ranging from clectrophysiology to diagnostic imaging. Since then, his research has further advanced towards the integration of physics-driven and data-driven analysis through the development of physics (physiology)-informed Al models for several medical applications. He is currently a Full Professor and Head of the Signal Processing Systems Division of the Electrical Engineering Department at TU/e. He has co-authored over 400 peer-reviewed publications, 14 patents, one book, and 15 book chapters. Prof. Mischi is a Board Member of the Imaging Section of the EAU, the Chairman of the IEEE EMBS Benelux Chapter, Member of the Safety Committee of the WFUMB, and the Secretary of the Dutch Society of Medical Ultrasound. He also serves as associate editor for IEEE T-UFFC, IEEE RBME, CMPB, Sensors, WFUMB Ultrasound Open, and IRBM. He was a recipient of the STW VIDI Grant (2009), the ERC Starting Grant (2011), the ERC Proof of Concept (2019), and the EIC Transition (2022) for his research on angiogenesis imaging by ultrasound technology. He is also the co-founder of two startup companies (Angiogenesis Analytics and HiPerMotion).