University Health Network

UHN's Laboratory Medicine Program operated across three primary diagnostic pathology sites (Toronto General Hospital, Princess Margaret Cancer Centre, Toronto Western Hospital) and 29 satellite locations in Ontario, processing nearly 1 million slides per year with a team of 70 pathologists. Traditional microscopy was time-consuming, labor-intensive, and subjective, and physical slide transportation across the distributed network created risks of specimen loss and limited cross-site collaboration.

UHN implemented digital pathology using high-throughput scanners (300–450 slide capacity) at main facilities and smaller scanners at satellite sites, backed by local server storage and integrated with existing LIS and HIS systems. A modified UK College of Pathologists validation protocol was adopted for regulatory compliance, and each pathologist was equipped with dual 4K monitors. The program was architected to support remote sign-out, teleconference consultations, workload redistribution across sites, and future AI-assisted diagnostics.

The system enabled pathologists to work remotely and share cases across all 29+ locations, reducing dependence on physical slide transport and enabling cross-site workload redistribution. Faster retrieval of archived slides and real-time expert consultations became feasible. The article cites published evidence that AI integration in digital pathology can improve breast cancer metastasis detection sensitivity to 89.4% compared to 74.5% for unassisted pathologists, though UHN-specific outcome metrics were not yet reported at the time of publication.

• A hybrid scanner approach (high-capacity units at main sites, low-capacity units at satellites) balances throughput needs with budget constraints across distributed laboratory networks.
• Local server storage was preferred over cloud for data sovereignty and long-term cost control, but storage costs grow linearly each year and must be budgeted prospectively.
• Change management, staff training, and upstream improvements to staining and glass slide quality are as critical to success as scanner selection — poor upstream quality directly increases rescan rates.