Perfusion MRI

This is Matthew Sun, Assistant Professor of Neurosurgery, brain tumor surgeon, and at UT Southwestern in Dallas, Texas.

Today, I have the pleasure of speaking with Dr. Alexandra Toldea and Dr. Riccardo Ludovichetti at University Hospital Zurich about their paper recently published in Neuro-Oncology Advances titled Prediction of Treatment Failure in Patients with Glioblastoma with Perfusion MRI and Molecular Biomarkers.

Alexandra and Ricardo, thank you for joining us today.

First, I'll just start out with an introductory question.

Your data is interesting, it is novel, and thought-provoking for sure.

I wanted to follow up with a few questions that really get into the details of the data and why it's so interesting, as well as about some of the methodological choices that you guys made that also made the results interesting.

So to start, I want to ask, your data show that higher pretreatment, RCBV, and K-trans were associated with treatment response, which is actually somewhat the opposite of what much of the perfusion MRI literature would suggest, where higher perfusion often signals a more vascular-aggressive and faster-growing tumor.

So could you walk us through how we might make sense of this finding biologically at the cellular level in your data? And maybe, in other words, do you think you're looking at a more treatable tumor subtype, a better drug and radiation delivery into a well-perfused tumor bed or something else entirely in your findings?

The device requires a multi-tissue segmentation map.

This map is derived from processing the patient's pre-treatment dynamic contrast enhanced magnetic resonance imaging, or DCE MRI.

And the document notes this map is actually a validated output of a previously cleared device called TumorSightViz.

Wait, so the workflow is that the MRI goes into one software, TumorSightViz, which generates a segmentation map, and then that map automatically flows into TumorSightRisk along with the clinical data? That sounds like a sophisticated pipeline.

Ardora AI Breast relies on a scanned histopathology whole slide image.

It analyzes tissue prepared from formalin-fixed, paraffin-embedded tissue and stained using hematoxylin and eosin stains.

In contrast, Tumor Site Risk uses DCE MRI data.

One is looking at pathology slides under a microscope, and the other is looking at radiological imaging.