United StatesHealthcare and life sciences2023 – ongoingMayo Clinic
AI analysis of Ki-67 breast cancer slides
Mayo Clinic reported a 0.97 correlation between AI Ki-67 scores and an existing test across 62 cases
Problem
Mayo Clinic needed to validate automated tumor detection and cell classification before using AI to score Ki-67 in breast cancer tissue. Its laboratory workflow also required manual adjustments when the model made errors.
Approach
Mayo Clinic and Aiforia researchers trained convolutional neural networks in Aiforia Create using scanned slides. They tested tumor detection and cell classification, including performance across scanners. The team adapted the software to its laboratory workflow and validated the model for clinical use.
Results
Takeaway
The laboratory retained manual correction of tumor regions. Validation covered scanner variation as well as agreement with an existing test.
Categories
- Size
- Large enterprise
- Industry
- Healthcare and life sciences
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Partner co-development
- Evidence
- Company's own claim
Sources
Source summaries
The joint research poster reported clinical validation of a breast cancer Ki-67 model. It gave a Pearson correlation of 0.97 against a previously validated test and described manual correction tools.
Full details are in the original sources.