AI-assisted CT analysis for colorectal imaging — bowel preparation assessment, polyp detection, segmentation & measurement.
Investigational software. Development and validation ongoing.
Structured assessment based on the Boston Bowel Preparation Scale (BBPS). BBPS scores each colon segment 0–3; total score = Right + Transverse + Left.
The AI identifies suspicious colorectal regions — polyps, masses, abnormal wall thickening and other suspicious findings — and provides visual localization with overlays for clinician review.
AI-assisted segmentation and lesion measurement support standardized clinical review:
Each segment (right, transverse, left) is scored 0–3; the total is the sum of the three segments.
| Total BBPS | Interpretation |
|---|---|
| 0–3 | Poor |
| 4–5 | Fair |
| 6–7 | Good |
| 8–9 | Excellent |
Polyps below 5 mm are harder to detect; flat lesions can produce minimal protrusion from the bowel wall.
Residual stool and fluid can mimic polyps, hide actual polyps and reduce image quality.
Folded or collapsed bowel segments can obscure lesions and make small abnormalities difficult to evaluate.
Small lesions occupying only part of a CT slice may blend into surrounding tissue or appear smaller.
Fits within the radiologist's existing workflow — no additional imaging procedure required.
Current focus: clinical evaluation and product refinement. Next step: evaluation on external (e.g. Japanese) datasets and workflows. Longer term: integration and commercialization.
A live demonstration of the tool interface is available on request; validation datasets and detailed performance results are shared under NDA.