Pathology, a key discipline in cancer diagnosis, has been based on the human eye for more than a century. Every day, thousands of pathologists observe colored tissues under a microscope, spot abnormalities, and count cells. Meticulous work today subject to strong demographic tensions. The arrival of artificial intelligence is reconfiguring this practice by automating certain stages of diagnosis to make reading more reproducible.
At the heart of this change, image analysis software based on deep learning models capable of identifying and quantifying cellular structures invisible to the untrained eye. These tools transform histological slides into digital data, allowing biomarkers or abnormalities to be identified with a high degree of precision. The objective is to increase the pathologist’s observation skills and make diagnoses more reliable.
“Our tools considerably facilitate the work of practitioners. Automatic detection and counting saves them valuable time,” explains Fanny Sockeel, co-founder and CEO of Primaa. The Parisian company has developed an AI platform capable of analyzing tissue images to detect breast and skin cancers. Its models, trained on large annotated datasets, recognize patterns associated with tumors and generate quantitative measurements useful for clinical decisions.
This approach reduces inter-observer variability, an issue often underestimated. Two pathologists may interpret the same slide differently, impacting classification and treatment. AI, by standardizing the analysis, introduces continuity into diagnostic practice. For health establishments, it also offers a partial response to the shortage of specialists, while reducing the time between sampling and result.
The gains are not limited to productivity. By systematizing quantification, these tools open the way to more predictive medicine. Current work focuses on the correlation between morphological signatures and clinical evolution: a shift towards the anticipation of relapses or the personalization of treatments. This convergence between digital pathology and statistical modeling could ultimately redefine the place of diagnosis in the care chain.
The transformation is nevertheless complex. The implementation of these systems in hospitals requires a digital infrastructure effort, regulatory validations and appropriate team training. Clinician confidence is based on the transparency of models and the traceability of algorithmic decisions. In Europe, regulations on AI-based medical devices impose scientific and ethical rigor which could become a competitive advantage.
Primaa is part of this dynamic, founded in 2018 by Fanny Sockeel, Stéphane Sockeel and Marie Sockeel, the startup has just raised 7 million euros with the MH Innov’ and Elaia partnership fundsof SWEN Capital Partnersof Super Capital and of members of the Wendel family. This funding will be used to strengthen the commercial and technical teams, accelerate the development of new AI modules, and prepare a FDA certification for its entry into the American market.
