Overview
Solen H&E is a computational pathology pipeline for routine hematoxylin and eosin (H&E) whole-slide images (WSI). It runs deep learning tissue segmentation (e.g. cancer, cancer-associated stroma, necrosis, artifacts) and cell detection and classification (e.g. cancer, fibroblasts, lymphocytes, plasmocytes, eosinophils, neutrophils). It then turns these into interpretable slide-level features covering tumor burden, cell composition, region-specific densities, nuclear morphology, and tumor-infiltrating lymphocyte (TIL) spatial distribution.
Outputs are structured, analysis-ready tables at three levels: slide-level features for cohort analysis, tile-level cell counts for heterogeneity mapping, and cell-level morphology (area, perimeter, circularity, cell type) for custom downstream work. Translational and biomarker teams can use them to profile the tumor immune contexture, stratify patients retrospectively, generate hypotheses for H&E-based biomarkers, and enrich clinical trial datasets from slides they already have, without additional staining.
The model deploys in your own AWS account through Amazon SageMaker, so slide images and results stay in your environment. Solen H&E is developed by Waiv. For Research Use Only. Not for use in diagnostic procedures.
Highlights
- Quantitative TME profiling from routine H&E: slide-level features covering cell composition, density, morphology and spatial organization, from slides you already have.
- Analysis-ready outputs at slide, tile and cell level: structured tables that plug straight into biomarker discovery and translational research workflows.
- Runs in your AWS account: deploys on Amazon SageMaker, so whole-slide images never leave your environment.
Details
Features and programs
Financing for AWS Marketplace purchases
Pricing
Vendor refund policy
All fees are non-refundable, except as otherwise provided in a written agreement between the customer and Waiv or as required by applicable law.
How can we make this page better?
Legal
Vendor terms and conditions
Content disclaimer
Delivery details
Amazon SageMaker algorithm
An Amazon SageMaker algorithm is a machine learning model that requires your training data to make predictions. Use the included training algorithm to generate your unique model artifact. Then deploy the model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Version release notes
New outputs
- Readouts now cover all 13 cell types the model detects, up from 6. The new ones are macrophages, muscle cells, endothelial cells, red blood cells, epithelial cells, apoptotic bodies and mitotic figures. Each gets counts, proportions, densities, per-tile count maps and morphology features.
- Minor Stromal Cell is no longer reported, so cell proportions now sum to 1.
Broader slide support
- The algorithm accepts any whole-slide format it can read, including DICOM WSI (one slide per folder), OME-TIFF and other formats that weren't supported before.
- A slide that can't be read no longer stops the whole job. Only that slide fails, and the reason is written to the output manifest.
- Slides whose base resolution is coarser than 1 µm/px are rejected, because results at that scale would be unreliable.
Faster processing
- Several stages run in parallel, so slides process faster on both ml.g6.4xlarge and ml.g4dn.4xlarge.
Privacy of job logs
- Training-job logs no longer print the internal pipeline configuration.
Unchanged
- Model weights, the input channel and the output layout stay the same. Existing integrations keep working without changes.
Additional details
Inputs
- Summary
This product runs as a SageMaker training job. It trains nothing: it applies pretrained models to the slides you provide and returns the analysis as the job's output.
Input: one training channel named "slides", pointing at an S3 prefix of H&E-stained whole-slide images scanned at 40x (about 0.25 micron per pixel). SageMaker copies the prefix onto the instance (File mode), so any folder structure works and files that are not slides are ignored. One job processes the whole prefix.
The real-time endpoint and the batch transform accept a JSON request but do not analyze images: they return a description of the product. Use a training job for the analysis.
- Input MIME type
- application/json
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
slides | Training channel: S3 prefix of the H&E whole-slide images to analyze. | File mode, application/octet-stream, no compression | Yes |
dataloader_profile | Hyperparameter. DataLoader tuning matched to the instance type. Leave unset to use the pipeline defaults. | ml_g6_4xlarge, ml_g4dn_4xlarge or ml_g5_xlarge | No |
num_workers | Hyperparameter. DataLoader worker processes; 0 is usually fastest. | Integer ≥ 0 | No |
Support
Vendor support
Email: support@wearewaiv.com
Waiv can provide technical support for deployment, input requirements and interpretation of Solen H&E outputs. Buyers under a separate agreement with Waiv receive the support terms defined in that agreement. This can include dedicated scientific support from Waiv's translational team for feature interpretation and custom analyses.
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.