To improve the specificity of MR imaging for diagnosing prostate cancer, Philips will incorporate its AI MR imaging applications into an integrated solution developed with AI software developer and imaging biomarker specialist Quibim, which will include its QP-Prostate platform.

Designed to automate prostate gland segmentation in real time in MR scans, QP-Prostate will provide quantitative insights to improve the specificity of exams, which are needed to better rule out negative results and save patients from paying additional costs for invasive and painful biopsies. The most pertinent quantitative biomarkers are the texture descriptors extracted from the T2 image and a normalized apparent diffusion coefficient to enhance the accuracy of region clustering and classification

Philips MR AI applications will improve the speed and quality of the results. They include its high-resolution T2w imaging; high-resolution diffusion-weighted imaging and quantitative processing; Smartspeed AI deep-learning package, which boosts resolution by 65% to improve image quality; and dynamic contrast-enhanced T1w package for permeability and perfusion imaging.
Combined with Quibim’s software, they will help detect clinically significant cancers earlier, drive personalized treatments, and improve first-time diagnoses. Additionally, the integrated application will be used alongside workflow enhancements to reduce the effects of staff shortages, burnout, and cost constraints that radiology and oncology departments face.

"The solution offers the capability to export regions in standardized formats, including DICOM-RT and DICOM-SEG, facilitating seamless interoperability with biopsy planning and radiation planning software that adhere to these industry standards. This interoperability enhances the integration of imaging data into the broader clinical workflow," Gwenael Herigault, clinical leader for MR at Philips, and Angel Alberich-Bayarri, CEO of Quibim, told HCB News.

The QP-Prostate platform creates structured reports with quantitative information based on PI-RADS v.2.1 guidelines. It automates prostate gland volume calculations and extracts clinical quantitative data through diffusion and perfusion analysis. Through convolutional neural networks, it automatically identifies prostate anatomy to segment each subregion and enhances image quality, and reduces heterogeneity with automatic registration, motion correction, and spatial smoothing features. Additionally, it supports both biparametric and multiparametric MR exams.

MRI Homepage