Artificial intelligence for prostate MR
OnQ Prostate (CorTechs Labs Inc., San Diego, CA, USA), Quantib Prostate (Quantib, Rotterdam, The Netherlands), Pi (Lucida Medical, Cambridge, UK), Syngo.via (Siemens Healthineers, Erlangen, Germany), ProstatIDTM (Bot Image Inc., Omaha, NE, USA) aim to improve prostate cancer detection and diagnosis using MR and AI.

Note that some products are interactive and allow for manual input from the user (Quantib, Plexo), and provide heat maps or DCE kinetic curve displays for user-drawn lesion segmentations. Others are fully-automated, have been tested for accurate volumetric measurements, and can operate in the background (OnQ, Pi, ProstatID). A key difference between OnQ and Pi and ProstatID is that OnQ relies on its patented advanced diffusion MR technique called Restriction Spectrum Imaging (RSI) which must be programmed into the acquisition sequences of the MR systems. RSI has shown an increase in the conspicuity of restricted water, an indication of aggressive cancer. ProstatID and PI both utilize the existing T2, DWI and ADC sequences available by the system manufacturers. Currently, neither are FDA-approved. Pi of Lucida Medical is not for sale in the USA whereas a training version of ProstatID, “QT Prostate MRI” is.
Finally, three of these products not only have the ability to automatically segment the prostate and provide volumetric measurements, but also automatically highlight regions suspicious for cancer, calculate their volumes, provide lesion risk scores and generate reports including 3D locator views of segmented lesions within the prostate and recommended PI-RADs scoring (Pi, Syngo.via, ProstatID). Syngo.via is only available on Siemens platforms, while ProstatID, QT Prostate MRI, and PI are available as a software-as-a-service (SaaS) and are MR system agnostic.

In conclusion, artificial intelligence has enormous potential in the field of radiology. We have provided examples of different software products utilizing varying degrees of AI for their implementation in prostate cancer MR that are currently available to assist radiologists. There are many applications where a program can mimic, augment, or even improve the decisions of the human. These applications vary from simple decision workflow programs to full-blown deep learning models. No matter the case, AI has something to offer radiology and, by extension, the patients in need of quality care.

About the author: Randy Jones is the CEO of Bot Image, Inc.

Back to HCB News

Artificial Intelligence Homepage