Endoscopy AI

Beyond polyp detection — characterization and surgical decision support

Reads colonoscopy video natively. Paris classification, NICE typing, pit pattern analysis, and non-lifting sign detection with auto-generated surgical briefs.

0fps

Real-Time Video Analysis

0

Classification Systems

Core Capabilities

The full endoscopic AI pipeline

Native Video Understanding

Reads colonoscopy video natively at the frame level. No frame sampling or snapshot extraction needed — the scope IS video, and our model treats it as such.

Paris Classification

Automated morphological classification of detected lesions using the Paris endoscopic classification system (0-Ip, 0-Is, 0-IIa, 0-IIb, 0-IIc, 0-III) for standardized reporting.

NICE Typing

Real-time NBI International Colorectal Endoscopic (NICE) classification to predict histology during the procedure, enabling optical diagnosis without biopsy.

Pit Pattern Analysis

Kudo pit pattern analysis using magnification chromoendoscopy features to differentiate neoplastic from non-neoplastic lesions at the mucosal level.

Non-Lifting Sign Detection

Automated detection of the non-lifting sign during saline lift, a critical indicator of submucosal invasion that determines endoscopic resectability.

Surgical Brief Generation

Auto-generated surgical briefs for colectomy referral when lesions exceed endoscopic resectability criteria, including lesion characterization, location, and staging data.

Clinical Pipeline

From detection to surgical decision in one procedure

01

Polyp Detection

Real-time polyp detection during live colonoscopy with bounding box overlay and confidence scoring, operating at native video frame rate.

02

Characterization

Paris classification, NICE typing, and pit pattern analysis to characterize each detected lesion morphologically and histologically.

03

Resectability Assessment

Non-lifting sign detection and submucosal invasion prediction to determine whether endoscopic resection is feasible or surgical referral is required.

04

Surgical Decision Support

Auto-generated surgical briefs with comprehensive lesion data, recommended approach, and colectomy referral documentation for the surgical team.

Beyond Detection

Where existing solutions stop, PixelMD begins

Current endoscopy AI products focus narrowly on polyp detection. PixelMD extends the clinical value chain from detection through characterization, resectability assessment, and automated surgical referral documentation.

Existing Solutions

Polyp detection
Bounding box overlay

PixelMD Endoscopy

Polyp detection + characterization
Paris, NICE, pit pattern classification
Non-lifting sign detection
Resectability prediction
Auto surgical brief generation

PixelMD

Endoscopy AI

Detect
Classify
Assess
Refer