99%
sensitivity
Micro Solutions / Life Science
AI module for analysis of colon histology KFBio AI Colon
AI analysis of colon biopsies: sensitivity 99%, specificity 95%
Product overview
- KFBio AI — Colon
- KFBio
- China
- KFBio AI Models
- Colon
AI analysis of colon biopsies: sensitivity 99%, specificity 95%
Heat maps and automatic gradation of neoplasia speed up viewing
An AI system for automated analysis of colon biopsies, cancer detection and classification of adenomatous lesions, heat maps and automatic gradation help reduce viewing time and increase reproducibility of conclusions.
The system detects adenocarcinoma, mucinous and ciliform cell carcinoma, tubular, villous and tubulous adenomas, dentate lesions and hyperplastic polyps. The algorithm segments pathological sites, separates low and high degrees of neoplasia, and generates a structured report for integration with LIS/PIS.
The model is trained on more than 20,000 slides from different parts of the colon and stages of the adenoma-carcinoma sequence. The manufacturer claims 99% sensitivity and 95% specificity according to multicenter validation data. Supports heat maps, quantitative assessment of the villus component and export of the structured result.
Key figures
95%
specificity
>20 000
learning-slide
Applications
- Gastrointestinal pathology
- Colorectal cancer screening
- Histological diagnosis of biopsies
Models and configurations
- Colon
Product details
AI analysis of colon biopsies
segmentation
Classification of lesions
adenomas and degrees of neoplasia
Analysis opportunities
- Detection of adenocarcinoma, mucinous and periform cell carcinoma
- Classification of tubular, villous and tubular adenomas
- Recognition of jagged lesions and hyperplastic polyps
- Separation of low and high degree of neoplasia
Accuracy and validation
- 99% sensitivity and 95% specificity according to multicenter validation
- Training on more than 20,000 slides from different parts of the colon
- Quantitative assessment of the villi component
Accountability
- Segmentation of pathological sites and heat maps
- Structured report for integration with LIS/PIS
Diagnostics
| Identified changes | adenocarcinoma, mucinous and ring-like cell carcinoma, tubular/villous/tubulous adenomas, dentate lesions, hyperplastic polyps |
|---|---|
| Gradation | low and high degree of neoplasia |
precision
| Sensitivity | 99% |
|---|---|
| Specificity | 95% |
| Training sample | over 20,000 slide |
Accountability
| Visualization | segmentation of sites, heat maps |
|---|---|
| Integration | LIS/PIS, Structured Report |
Related equipment types
- Digital slide scanners
- Digital slide management system
- Histological processors of tissues
- Microtomes
Related products
- KFBio KF-DSMS — Digital slide management system
- KFBio KFBio AI — Stomach — Histological processors of tissues
- Medite S180 — Microtomes
Classification
- Histology and cytology
- AI analysis of histological preparations of the colon
- Histological and pathomorphological equipment
- Digital pathology
- Viewing, archiving and telepathology, AI modules
- KFBio AI — Colon