AI System for Preliminary Quality Classification of Digital Pathology Images

The problem

In day-to-day operations, scanning defects can delay specialist review. When the slide image, scanner metadata, barcode and quality protocol are scattered across different screens, documents or team records, problems do not become visible in time.

Fragmented data delays preparation of the slide quality review queue and forces staff to search manually for supporting evidence again. Before making a decision, the pathology technician and pathologist have to complete missing fields and verify inconsistencies one by one.

A common exception is that a valid specimen with limited material may fall below standard coverage. Incomplete or incorrect automation that overlooks this context could lead to a valid specimen being rescanned or an inadequate scan being accepted, so the output cannot be applied directly.

What we set out to improve

• Rate of confirmed rescan candidates
• Rate of outputs corrected or rejected by the pathology technician and pathologist
• Time between the source record and the review output

How the system works

1. 01 Define the scope of the review and the authorised pathology technician and pathologist roles
2. 02 Collect the inputs: slide image, scanner metadata, barcode and quality protocol
3. 03 Check source identity, date, version and data freshness
4. 04 Check barcode and file integrity
5. 05 Classify focus and coverage signals
6. 06 Check the exception: a valid specimen with limited material may fall below standard coverage
7. 07 Place low-confidence or inconsistent records in a separate human review queue
8. 08 Write the source identity, rule, output and time to the audit trail
9. 09 Have the pathology technician and pathologist approve, correct, defer or reject the output

Methods we used

• A technical image quality model that does not classify pathological findings
• Source, format, version and required-field checks for the slide image, scanner metadata, barcode and quality protocol
• A missing-data warning instead of an estimate when inputs are missing or inconsistent
• Checking exceptions separately from the main rule
• Storing the source identity, rule result, slide quality review queue and human decision together

Where people stay involved

Data and security

Access to the slide image, scanner metadata, barcode and quality protocol is limited to the minimum permissions required for each role. Identifiers and sensitive fields are masked where possible, and only the necessary section is sent to the model rather than the entire raw file. Access, outputs and decisions by the pathology technician and pathologist are recorded; the system does not enter production until the organisation has approved the retention period and data location.

Who this suits

Frequently asked questions