Luis Cano MD PhD is a Paris-based expert on digital pathology. Find his Linked in here. His article feed here. His Substack articles, "Beyond the Slide," here.
His new 5000-word mega-essay about digital pathology system dynamics:
https://beyondtheslide.substack.com/p/the-invisible-war-in-digital-pathology
Enjoy the original essay but here's a 300 word summary (AI):
Digital pathology, Cano argues, is less a neat “digital upgrade” and more an invisible war: a messy, protracted collision between technology that already works and systems that are nowhere near ready to absorb it. The microscope era was stable, tactile, and culturally entrenched; digital workflows turn glass into data, demanding scanners, clouds, networks, standards, and cybersecurity.
COVID accelerated adoption by necessity, proving feasibility but also exposing fragile infrastructure and improvisational workflows, especially in low-resource settings where digital tools could help most but basic prerequisites—power, bandwidth, capital—are lacking. The result is a tug-of-war between an analog world that won’t die and a digital one not yet fully born.
The conflict is driven less by algorithms than by misaligned incentives and timelines. Startups sprint on 18-month venture clocks; hospitals move on 5–10-year cycles; pathologists, as final signatories, face an identity and liability crisis.
Economics form a central minefield: capital costs are huge, ROI is mostly indirect efficiency, and reimbursement for digitization or AI is nascent and fragmented. Technologically, proprietary formats, weak interoperability, domain shift, and brittle real-world performance keep AI stuck in “demo mode.” Regulatory fog deepens the tension: the FDA’s PCCP and the EU AI Act try to tame evolving, data-hungry models with frameworks built for static devices, while liability for AI-driven errors remains unclear.
Cano’s core thesis is that progress hinges on alignment, not more clever models. He points to federated and swarm learning, shared infrastructures like BigPicture, and human-centered design as early blueprints for cooperation. The future pathologist becomes an integrator of multi-modal information, not a human scanner; AI should amplify rather than replace. The “war” ends, he suggests, when stakeholders stop behaving like rival armies and start acting as architects—building standards, economic models, regulations, and lab designs that are interoperable, explainable, and explicitly human-centered.
