AI Ecosystem

SatSight Dx

Smarter MSS detection. Streamlined pathology workflows.

SatSight Dx flags microsatellite-stable colorectal cancer cases from the H&E slide that already exists. An ensemble of 100 AI models, validated across five independent international cohorts.

Try the live demo Request integration
Research use only · not for primary diagnostic use
SatSight Dx · Inference
Problem & Solution

Most CRC cases are MSS.
We test all of them anyway.

MSI/MMR evaluation is mandated on every colorectal resection. SatSight Dx scores the H&E slide first, so molecular workup is reserved for the cases where the result actually changes care.

The problem

Reflex molecular on every CRC

Today, every colorectal resection is routed to MSI-PCR or MMR-IHC. Most cases turn out microsatellite-stable after weeks of reagents, technologist time and pathologist scoring.

  • 100%of cases tested molecularly
  • ~70%turn out MSS
  • 5 to 15 daystypical PCR / IHC turnaround
  • Highreagent & pathologist load
Our solution

AI triage on the H&E that already exists

SatSight Dx reads the same H&E slide pathologists already cut. An ensemble of 100 AI models returns a confidence-aware call, so molecular testing concentrates on MSI and Undetermined cases.

  • 59 to 78%of slides safely triaged
  • <2%MSS mis-flagged as MSI
  • Same sessionresult on the existing H&E
  • 0extra reagents or staining
The shift

From three weeks
to less than sixty minutes.
On the slide that already exists.

SatSight Dx · CRC MSI Classifier
Try it · Live demo

Try it on a slide.
See the prediction.

Example anonymized H&E cases below, for demo purposes only. Switch between original, tissue segmentation and attention heatmap. Generate a sample report to see what gets delivered.

How it works

One slide in.
A confidence-aware call.

  1. 01

    Drop the H&E

    Drag-and-drop a whole-slide image, or run SatSight Dx natively inside Roche Digital Pathology or Morphle Lab environments. No new hardware, no extra staining.

  2. 02

    100-model ensemble

    Each slide is scored by an ensemble of 100 AI models. The mean MSI probability and 95% confidence interval surface MSS, Undetermined, and MSI cases.

  3. 03

    Pathologist-ready report

    Every report includes classification, attention heatmap and tissue segmentation across tumor, stroma, mucus, lymphocytes and normal epithelium.

>95%AUC · 5 independent cohorts
>97%Sensitivity · all validation cohorts
<2%MSS mis-flagged as MSI
Who it's for

Built for four
customer segments.

The same model, four operating modes. Choose the one that fits how your team already works.

Pathology labs

API integration: whatever scanner you use, connect to our API and our AI runs SatSight Dx to detect MSI / MSS.

  • Reflex MSI from routine H&E
  • Roche Digital Pathology · Morphle Lab integration

Pharma & clinical trials

Stratify enrollment for immunotherapy and run retrospective biomarker analysis on archive cohorts.

  • Validated on 5 independent international cohorts
  • Diverse staining protocols and patient populations

Diagnostics & IVD

Benchmark in-house MSI assays against an uncertainty-aware AI baseline trained on routine H&E.

  • Ensemble of 100 models with confidence intervals
  • AUC > 0.95 across external cohorts

Research & AMCs

Pair classification with morphology, segmentation and attention overlays to surface spatial patterns worth chasing.

  • Attention heatmaps
  • Tissue segmentation: tumor · stroma · mucus · lymphocytes · normal epithelium
Get SatSight Dx

Bring an H&E.
We'll bring the model and the result.

Speak with the PAICON team to learn how SatSight Dx can be deployed in your workflow environment.

Let's connect