Cancer imaging

Multiparametric MRI for prostate cancer imaging

Prostate cancer is the second most common cancer in men worldwide. Prostate cancer often progresses slowly and may not cause noticeable symptoms in its early stages. However, when left untreated, it can become aggressive and spread to other parts of the body. Multiparametric Magnetic Resonance Imaging (mpMRI) has emerged as a powerful tool in the diagnosis and management of prostate cancer. This advanced imaging technique combines multiple MRI sequences to provide detailed and comprehensive information about the prostate gland. It combines a T2-weighted sequence for detailed anatomical images of the prostate, with Diffusion-Weighted Imaging (DWI), evaluating the movement of water molecules in tissues, and Dynamic contrast-Enhanced Imaging (DCE-MRI), which is based on injection of a contrast agent to probe tumor perfusion and vascular permeability. The latter is particularly interesting to assess angiogenesis, an hallmark of cancer growth, which results in the formation of an (abnormal) vascular network feeding the tumor.

To provide a quantitative method to analyze DCE-MRI, we developed Magnetic Resonance Dispersion Imaging (MRDI), a method providing simultaneous assessment of the vascular architecture and permeability, with several advantages compared to well-known analysis by Tofts’ model. MRDI provides interpretable, quantitative features for cancer diagnostics.

To boost the diagnostic accuracy, we propose a radiomic approach combining features extracted from MRDI with texture analysis of all three mpMRI series. Additionally, we seek a link between the imaging features extracted from MRI and the genomics/transcriptomics of prostate cancer tissue by a radiogenomic analysis. This provides a basis to possibly infer genomic markers from non-invasive imaging in the future.

Quantitative contrast-enhanced ultrasound

Contrast-enhanced ultrasound (CEUS) is a powerful diagnostic tool provide exquisite characterization of the vasculature by the injection of ultrasound contrast agents. These are composed by microbubbles with a gas core encapsulated by a shell. Because of their non linear response, their backscatter ultrasound intensity can be distinguished from that of surrounding tissues, using clever image formation and processing strategies. Over the years, within the BM/d group, several methods have been developed for quantitative analysis of CEUS loops, ranging from dispersion modeling of the contrast agents transport, to spatiotemporal analysis by different similarity measures and system identification strategies. All these quantification techniques are referred to as contrast ultrasound dispersion imaging (CUDI).

This line of research focuses on further improving CEUS quantification, and optimally combining CUDI with other quantitative imaging strategies (e.g texture analysis, shear wave elastography) by interpretable machine learning to improve diagnosis of several solid tumors.

Ultrasound molecular imaging

While microbubbles used for standard CEUS enable exquisite visualization of the vasculature, it lacks the specificity of molecular contrast agents. By decorating the microbubbles shell with targeting ligands, ultrasound molecular imaging has become possible in the past decade. These novel agents can bind to molecular targets on the vessel wall, thus providing specific enhancement where the molecular targets is over-expressed. In this line of research, we adopt a pharmacokinetic modeling approach to describe the transport and accumulation of these novel contrast agents, enabling the extraction of quantitative parameters related to binding, which can be used for diagnosis and therapy monitoring.

The pharmacokinetic approach previously adopted for MRDI and to describe the transport of targeted microbubbles, is further extended to describe the more complex kinetics and distributions of nanobubble ultrasound contrast agents. Being about 10 times smaller than conventional microbubbles, these novel agents exhibit deeper penetration in the capillary network and may cross the vascular endothelium, making them suitable to reach targets beyond the vasculature. One example is the prostate-specific membrane antigen (PSMA), which is over-expressed on prostate cancer cells. In this line of research, we extract parameters related to extravasation and binding to characterize tumor tissue.