About sMRD

sMRD (spatial Molecular Residual Disease) is an innovative, tissue-based, next-generation sequencing approach applied to tumor tissue, leveraging a novel tumor bed reconstruction method to define 3D spatial patterns associated with relapse. This online resource accompanies the sMRD manuscript by Ransohoff, et al. from Stanford University. We invite users to explore renders and relapse predictions of their own clinical cases or to explore those from our study.

Why It Matters

Defining metastatic potential in the primary tumor bed and improving relapse predictions

Relapse Prediction

Identifying the high risk spatial dispersion pattern linked to metastases

Clinical Integration

Leveraging routinely-collected clinical data

Precision Oncology

Toward risk-adapted precision adjuvant treatments

Research Foundation

Integrating MRD detection and spatial pathology to characterize spatial response patterns

Frequently Asked Questions

Find answers to common questions about sMRD technology, image requirements, data privacy, and getting started with the platform.

Image Requirements & Preparation

You may use any macroscopic image that captures the spatial registration of your tissue slabs. Images with a pure black background work best, as they simplify tissue detection and improve downstream rendering. If your images are brightfield with a non-uniform background, we have had good results using tools such as Adobe Photoshop to automatically detect the background and render it black prior to upload.

Images do not need to be pre-scaled, but scale information is strongly recommended. Best practice: - Know the pixel size of your images when possible - Use similar pixel resolution across all images within the same case To support this, the Annotate module includes a built-in tool that can derive pixel resolution from any known real-world distance visible in the image (for example, a ruler or labeled measurement).

- Rotated images are supported and can be corrected in the Stack module - Mirrored images are not currently supported If mirroring is required for your workflow, please contact us to discuss potential extensions.

The platform currently supports: - .jpeg - .jpg - .png Each image must be 50 MB or smaller. If you require support for additional formats, feel free to contact us.

Image Annotation & Naming Conventions

Using separate images is recommended, but not mandatory. The platform supports uploading two identical images (same dimensions and tissue alignment): - Raw image: {case_id}_{image_nr}_r.jpeg example: 34_1_r.jpeg - Annotated image: {case_id}_{image_nr}_a.jpeg example: 34_1_a.jpeg Note that case_id and image_nr may only contain integer numeric values. (SH-34 is not allowed, the integer numeric part 34 is allowed) If only a single image is uploaded and it contains annotations, those annotations may remain visible in the final render.

Following the naming convention is strongly recommended for smooth ingestion. The upload interface includes user input fields that allow you to correct or override naming inconsistencies if needed. Reminder: - Raw image: {case_id}_{image_nr}_r.jpeg example: 34_1_r.jpeg - Annotated image: {case_id}_{image_nr}_a.jpeg example: 34_1_a.jpeg Note that case_id and image_nr may only contain integer numeric values. (SH-34 is not allowed, the integer numeric part 34 is allowed)

Case Size & Upload Limits

There is no formal upper limit on the number of tissue slabs or images per case. Fair usage is expected, as extremely large uploads may affect performance or processing time.

Data Storage, Privacy & Compliance

Yes. Uploaded images are stored on secure Stanford servers. - Data is siloed per user - You can only access your own cases - You may delete your case data at any time via the website

Users should not upload content that may qualify as PHI under applicable regulations. Please ensure that: - Case IDs are fully de-identified - Case descriptions do not contain patient identifiers - Image metadata (EXIF tags) are removed prior to upload

Scope, Validation & Intended Use

Yes. - The 3D reconstruction pipeline works for any tissue type - Pathologist ledger metrics (e.g. dispersion score, RCB) are currently optimized for neoadjuvant-treated breast cancer

No. sMRD is currently for research use only and has not yet been clinically validated.

Support & Feedback

Please contact: cll@stanford.edu