Aiforia Technologies is likely feeling an apt sense of euphoria after racking up its third CE mark in less than a year for another of its artificial intelligence models to aid in cancer diagnoses.
The Finnish startup’s AI tools are designed to analyze tissue images in the pathology lab, looking for biological changes associated with cancer.
Using the AI models to detect and assess these biomarkers “can lead to significant time savings and clinical workflow improvements for pathology labs, as the evaluation time per case required from a pathologist is reduced,” according to Juuso Juhila, Ph.D., Aiforia’s director of clinical products.
The latest of Aiforia’s models to receive European approval seeks out estrogen receptors, which are present in about 80% of breast cancers. ER-positive breast cancers are more likely to respond to highly effective hormone therapies like endocrine therapy, Juhila said.
The new model examines whole slide images of potentially cancerous tissue. It automatically identifies tumors and calculates the percentages of ER-positive and ER-negative cells present in the sample. It can also point out “hotspots” that are particularly crowded with estrogen receptors.
With that analysis, pathologists can make a definitive cancer diagnosis more quickly. Physicians also get a head start on treatment planning, since knowledge of a patient’s estrogen receptor count can help direct them toward or away from hormone therapy.
“We believe that AI and digital solutions are the future of cancer diagnostics," said Aiforia CEO Jukka Tapaninen. "Our aim is to help alleviate the challenges faced by global healthcare systems, to help reduce costs and to ultimately support pathologists in their work to improve patient outcomes and enable precision diagnostics."
The ER calculation tool joins Aiforia’s slate of two other CE marked AI pathology models, both of which earned their European approvals in recent months.
The first arrived last summer and is also aimed at improving breast cancer diagnostics. Rather than analyzing estrogen receptors, however, that model sifts through whole slide images to tally up cells that are either positive or negative for Ki67, a protein that’s associated with the rapid growth and division of cancer cells.
Just a few months later, in November, Aiforia scored its second CE mark, this time for a diagnostic aid for lung cancer. That AI tool analyzes PD-L1, another biomarker that can demonstrate how aggressively cancer is spreading. If the model detects a high percentage of PD-L1 in a patient’s cells, it can indicate that they’d be a promising candidate for checkpoint inhibitor therapy, a form of immunotherapy.
Aiforia’s work to develop these and other pathology models is fueled by a recent outpouring of financial support that brought in 17.5 million euros, or about $19 million U.S. in series B funding.