AI image analysis detects organ damage in prostate cancer therapy

In a recent study, researchers from the Departments of Radiology and Nuclear Medicine at TUM University Hospital examined medical records and imaging data from 121 patients treated with lutetium-177 PSMA. Their goal was to find a way to predict kidney impairment without the need for invasive tissue sampling. 

“In an earlier study, we found that patients whose kidney function worsened after lutetium-177 PSMA therapy showed changes in kidney structure,” explains lead author Dr. Lisa Steinhelfer. “Since it's not feasible to routinely take tissue samples, we wanted to explore whether these changes could be detected using less invasive methods.” 

The researchers analyzed routinely collected CT scans and blood values to identify indicators of organ damage that occur before functional impairments become measurable. 

While patient age or kidney length showed no predictive value, changes in kidney volume proved significant: 

• A 10 percent or greater decrease in kidney volume within six months of therapy onset correlated strongly with a significant decline in renal function within the following six months. 

• These volume reductions are often so minor that they go unnoticed in conventional assessments. 

“These changes in kidney volume are very subtle. They can easily be missed during routine image assessments because clinicians are mainly focused on tracking tumors and other critical findings,” says Prof. Matthias Eiber, senior author of the study, together with Prof. Rickmer Braren. 

“In contrast, image analysis algorithms – if properly trained – can reliably detect even these minor changes,” adds Dr. Friederike Jungmann, first author with Dr. Steinhelfer. 

The research team previously showed that volume changes in the spleen may signal early disruptions in blood cell production. 

“Many cancer therapies can affect liver function or the hematopoietic system. I believe our approach could help identify a wide range of treatment-related side effects much earlier than currently possible,” says Steinhelfer. 

MEDICA-tradefair.com; Source: Technical University of Munich