Mindpeak logo
Academic Research · Data Labeling

Mindpeak

Mindpeak is a German digital pathology company whose AI reads scanned tissue slides to quantify cancer biomarkers such as HER2, PD-L1 and Ki-67, plugging into the pathology platforms that diagnostic laboratories and biopharma research teams already run.

Active GDPR compliant Contact Sales No public API Verified by Guidaio
Overview

What is Mindpeak?

Mindpeak is a Hamburg company that applies artificial intelligence to digitised tissue slides, the raw material of cancer pathology. Its entire catalogue rests on one proprietary, patented foundation model called peakFrontier, which the company says was built on five million expert annotations over seven years of validation. The model works in four stages: it accepts an image in multiplex immunofluorescence, immunohistochemistry or H&E, segments the tissue into twelve clinically relevant classes, classifies individual cells into thirteen types, then segments nucleus, cytoplasm and membrane so that biomarker expression can actually be measured.

Two audiences are served from that base. For diagnostic laboratories, Mindpeak sells scoring suites organised by biomarker family: a breast suite covering HER2, Ki-67, ER, PR, PD-L1, H&E and p53; a pan-tumour PD-L1 suite spanning lung, gastric, breast, oesophageal, urothelial and ovarian indications; a pan-tumour HER2 suite aimed squarely at the HER2-low and ultralow ranges; and separate tools for prostate Gleason grading, onychomycosis screening and lymphocyte counting. peakDissect proposes the best regions to macro-dissect before molecular sequencing. For pharmaceutical companies, peakDiscover extracts spatial signatures from the tumour microenvironment, peakPredict supports patient stratification and trial endpoints, peakDeploy handles worldwide rollout, and peakAcademy trains pathologists on a sponsor's biomarker. PhenoScout AI is the code-free multiplex analysis product built for research teams.

Regulatory status varies product by product and has to be read carefully. Annex 1 of the published general terms lists twenty-one algorithms: nine carry CE-IVD marking in the European Union, and none is cleared by the FDA or authorised in Canada or the United Kingdom. Everything else is Research Use Only, and the terms restrict users in the United States to research purposes. Mindpeak states that its quality management system is certified to ISO 13485.

The software is rarely bought as a standalone product. It is designed to sit inside the digital pathology platform a laboratory already runs, or behind an interface Mindpeak provides. There is no public price list and no self-service sign-up: an engagement begins with a demonstration and ends in a signed order form.

What it does

  • Score immunohistochemistry biomarkers on a whole slide without a single click, covering HER2, PD-L1, Ki-67, ER, PR and p53
  • Detect and segment tumour tissue directly on routine H&E slides, including invasive breast carcinoma
  • Classify individual cells into thirteen types and segment nucleus, cytoplasm and membrane for accurate quantification
  • Analyse multiplex immunofluorescence images, including cyclic stainings of more than twenty channels
  • Propose the optimal regions for macro-dissection before molecular sequencing, preserving scarce tissue
  • Grade prostate biopsies by Gleason pattern and quantify tumour-infiltrating lymphocytes on H&E
  • Train pathologists on scored cases measured against verified ground truth, with exportable performance dashboards
Audience

When to use Mindpeak / When not to

A quick filter to help you decide if Mindpeak is the right fit.

When to use Mindpeak

  • High-volume anatomic pathology laboratories that want immunohistochemistry scored the same way on every case
  • Laboratory networks already running a digital pathology platform such as Proscia Concentriq, Pathomation or Techcyte Fusion
  • Biopharma discovery teams profiling the tumour microenvironment across multiplex immunofluorescence, IHC and H&E
  • Clinical development groups trying to cut inter-reader variability across multi-site trials and companion diagnostic programmes
  • Medical affairs teams that need to train pathologists worldwide on a specific biomarker before a therapy launches

When not to use Mindpeak

  • Patients and members of the public: this is a business-to-business product sold to institutions through a signed order form
  • Laboratories in the United States looking for a clinical-grade tool, since the terms restrict use there to research only and no product is FDA-cleared
  • Teams in Canada or the United Kingdom that need a regulatory-approved assay, as the published annex shows no authorisation in either country
  • Buyers who want a published price, a free trial or an online sign-up, none of which exist here
  • Organisations that cannot accept a perpetual licence over the images they upload, including their use to retrain the vendor's AI
Get started

How to use Mindpeak

A typical end-to-end flow, from setup to results.

  1. Start from the contact page: there is no online sign-up, so the entry point is the enquiry form or a booked demonstration slot
  2. Fill in your name, work email, position and whether you want to speak to the BioPharma or the Diagnostics team
  3. Discuss scope with Mindpeak, including which algorithms you need and their regulatory status in your country
  4. Check whether your existing digital pathology platform is already a Mindpeak partner, which removes most of the integration work
  5. Agree an order form setting out the services, the annual quota of image scans and the fixed price
  6. Choose a deployment mode: cloud-native browser access, or an on-premise dockerised installation inside your own network on request
  7. Connect your scanners, laboratory information system and image storage; no rip-and-replace is expected
  8. Verify that your slides meet the staining, resolution and file format requirements set out in the manual
  9. Open a slide in your usual viewer and read the biomarker score, which appears without further clicks
  10. Review the single-cell output before signing off, since final diagnostic responsibility stays with the clinician
Quick read

Pros & Cons

Pros

  • Regulatory status is published algorithm by algorithm in a freely downloadable annex, which is unusually transparent for this market
  • The quality management system is certified to ISO 13485, announced in its own press release rather than merely claimed on a badge
  • The AI slots into pathology platforms a laboratory already owns, so there is no rip-and-replace and little integration overhead
  • Deployment is flexible: cloud access through a browser, or a dockerised on-premise install inside the customer's own network on request
  • No bioinformatics skills are needed, since the models are pre-trained and run on upload with no pipeline to build
  • A substantial body of validation work is published, including posters at ASCO in 2023, 2024 and 2025 and multi-centre studies
  • The full general terms are available as a public PDF, so the contract can be read before any sales conversation

Cons

  • No price is published anywhere, and there is no free plan, no free trial and no online sign-up
  • The licence granted over uploaded images is perpetual and explicitly covers retraining the vendor's AI, with no documented way to opt out
  • Publishing research based on the results requires Mindpeak's prior written consent, which is a real constraint for academic laboratories
  • Most of the catalogue is Research Use Only: only nine of the twenty-one listed products carry CE-IVD marking, and none is cleared in the US, Canada or the UK
  • No customer-facing data processing agreement is published, even though the product handles patient tissue images
  • No hosting country or region is stated anywhere for customer data
  • The site contradicts itself in places, advertising eleven CE-IVD products where the terms annex lists nine, and both eighteen and nineteen platform partners in the same section
Pricing

Pricing & Plans

No price is published. Mindpeak operates no pricing page, and the usual paths return a clean error rather than a hidden one. There is no free plan and no free trial; the only route in is a demonstration or a contact request. The commercial mechanism is nonetheless documented in the general terms: fees are set in an order form and fall due within fourteen days, and they do not depend on the number of users on an account. What is billed is the number of annual requests sent for image analysis, with an agreed annual quota of scans included in a fixed price and any scan beyond that quota charged separately. Because that unit has no equivalent in this directory's billing vocabulary, the amount, currency and billing unit are left empty rather than approximated.

Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Mindpeak handles your data.

GDPR overview

GDPR treatment is concrete rather than declarative. Mindpeak GmbH names itself as controller at the top of a bilingual English and German privacy policy, cites its legal bases article by article, and details the full set of data subject rights including the right to object under Article 21. Being established in Hamburg, it needs no Article 27 representative and designates none. It has appointed an external data protection officer, ARTANA Digital GmbH, reachable at dataprotection@mindpeak.ai. Four processors are named outright, with standard contractual clauses covering transfers outside the EU. The company also states plainly in its FAQ that it is GDPR-compliant and made in Germany. One gap is worth noting: no customer-facing data processing agreement is published or offered, despite the product handling patient tissue images.

Who owns the data?

Mindpeak keeps everything on its side. Its terms state that it retains all intellectual property in the services, in any improvement to them and in any technology developed alongside them; the customer receives only a non-exclusive, non-transferable and time-limited right of access. Results may be used for the customer's own internal business and research purposes, but they may not be published without Mindpeak's prior written consent, and they may not be used to train or develop a competing artificial intelligence. Uploaded images remain the customer's responsibility: the customer warrants owning all rights to them and having obtained every necessary consent. Once the contract ends, access to past results through the services stops.

Reuse rights

Reuse is asymmetric, and the asymmetry is written down. On the customer's side, results may be used internally without asking, but publishing anything based on them requires Mindpeak's prior written consent, and using them to train or develop one's own artificial intelligence is forbidden outright. On the vendor's side, the Image Upload Policy and clause 7.1 of the terms are far broader: submitting an image for analysis grants Mindpeak and its affiliates a non-exclusive, royalty-free, perpetual licence, unrestricted in territory, duration and content, to copy, transmit, reproduce, publicly display, edit, reformat and sub-license that image. The stated purposes include the vendor's documentation duties, service delivery, advertising and product improvement, and name the retraining of artificial intelligence explicitly. No mechanism to withdraw images from that retraining is documented anywhere on the site.

Data retention & training

Retention summary
The published retention rules cover the website rather than the product. Data is erased once it is no longer needed for the purpose it was collected for, with one figure given: twelve months for fraud prevention and support improvement. Session cookies are deleted when you leave the site, IP addresses are kept for the session and then logged, and data may be held longer where European or national law requires it or where a contract must be performed. For the product itself there is an asymmetry worth knowing: the terms end your access to results when the contract ends, but the licence Mindpeak takes over uploaded images is expressly perpetual. No retention period is published for patient images or analysis results.
Trains on customer data
Yes
Subprocessors disclosed
Yes

Hosting summary

No hosting country or region is disclosed for customer data. The general terms say only that the services are made available on a central server or several servers, without naming a location, and neither the website nor the privacy policy fills that gap for patient tissue images. What the privacy policy does cover is the marketing website: it names Usercentrics for consent management, Dropbox International Unlimited Company in Dublin for hosting capacity, Google Ireland for website services and Spline in Delaware for 3D content, and it warns that onward processors may handle data in the United States under standard contractual clauses, with Dropbox additionally certified under the EU-US Data Privacy Framework. Those arrangements concern the website, not the product, and should not be read as the hosting location of clinical images. The one useful lever is deployment: an on-premise dockerised installation inside the customer's own network is available on request, which keeps images and results in-house. Anyone with a data residency requirement should settle it in writing before signing.

Availability

Where Mindpeak works

Country-level availability.

Not available in

United States clinical use excluded: the general terms limit the software and its user interface to Research Use Only there, and no product is FDA ClearedCanada not authorised for diagnostic use: the published regulatory annex marks all twenty One listed products as unapproved in CanadaUnited Kingdom not authorised for diagnostic use: the published regulatory annex marks all twenty One listed products as unapproved in the UK
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Mindpeak.

  • The perpetual, sub-licensable licence over uploaded images covers advertising and AI retraining, and should be reviewed by legal counsel and your data protection officer before any slide is sent
  • The compliance burden for patient consent sits with you: the terms require you to warrant that you own all rights to the images and have obtained every necessary consent
  • No customer-facing data processing agreement is published, which is a notable gap for a service handling patient tissue images
  • Check the regulatory status of each individual algorithm in the terms annex before any clinical use, because most of the catalogue is Research Use Only
  • Automation bias is a real hazard: the vendor positions the tool as decision support and states that final diagnostic responsibility remains with the clinician, so scores should be reviewed rather than accepted
  • The terms disclaim responsibility for the accuracy of diagnostic results, which shifts the clinical risk back to the laboratory
  • Certifications and CE marking belong to Mindpeak, but the hospital and laboratory logos shown on the site are clinical partners and customer references, not indicators of the vendor's own accreditation
Setup

Setup & Integrations

Technical difficulty

Low to moderate, and it depends on where you start. If your laboratory already runs a partner pathology platform, the vendor promises no extra development or technical overhead, with no rip-and-replace and no downtime. Models are pre-trained and run on upload, so no image analysis or bioinformatics expertise is required. Effort rises for an on-premise dockerised deployment, and the terms are explicit that interface programming, installation and configuration are not included unless separately agreed. Slides must also meet documented staining, resolution and file format requirements.

Deployment

Web appPlugin

Integrations

Proscia Concentriq Techcyte Fusion Pathomation ZEISS Axioscan 7 RareCyte Orion Akoya PhenoImager HT Lunaphore COMET
Company

Behind Mindpeak

Company name
Mindpeak GmbH
Founded
24/08/2018
Country of origin
🇩🇩 Germany
Headquarters
Zirkusweg 2, 20359 Hamburg, Germany
US office
Mindpeak Inc, Cambridge Innovation Center, 245 Main Street, 2nd floor, Cambridge, MA 02142, United States
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇱🇺 Luxembourg

Fundraising

October 2024 - USD 15.3 million Series A round led by Zeiss Ventures and InnoVentureFund, with participation from AI.FUND and the European Innovation Council Fund
June 2020 - approximately EUR 2 million in grant funding from the European Commission's Horizon 2020 programme and the Profi Transfer scheme of the Hamburg Investment and Development Bank

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement Mindpeak.

V Visiopharm
FAQ

Frequently asked questions

What does Mindpeak actually do?
It applies artificial intelligence to digitised tissue slides in order to quantify cancer biomarkers such as HER2, PD-L1 and Ki-67, and to detect tumour tissue on routine H&E. The results support both clinical diagnostics and pharmaceutical drug development.
Who is behind the product?
Mindpeak GmbH, Zirkusweg 2, 20359 Hamburg, Germany, registered at the Hamburg local court under HRB 151143 with VAT number DE318495606 and managed by Felix Faber, who founded the company with Dr Tobias Lang. A US subsidiary, Mindpeak Inc, operates from the Cambridge Innovation Center in Cambridge, Massachusetts.
Is the software approved for clinical diagnosis?
It depends entirely on the product. Annex 1 of the published general terms lists twenty-one algorithms: nine hold CE-IVD marking in the European Union, and none is cleared by the FDA or authorised in Canada or the United Kingdom. Everything else is Research Use Only, and users in the United States are restricted to research purposes.
How much does it cost?
No price is published. Fees are agreed in an order form and are based on an annual quota of image scans rather than on the number of users; scans beyond the quota are billed separately. There is no free plan and no free trial.
Will my slide images be used to train the AI?
Yes, and the terms say so plainly. Submitting an image grants Mindpeak and its affiliates a perpetual, royalty-free, sub-licensable licence to use it, and the stated purposes include advertising, product improvement and specifically the retraining of artificial intelligence. No opt-out mechanism is documented.
Can I publish research based on the results?
Not without permission. The general terms forbid publishing articles or research based on results generated by the software without Mindpeak's prior written consent, and they also forbid using the results to train a competing AI.
Can it run inside our own infrastructure?
Yes. The research product is cloud-native and browser-based by default, but a dockerised on-premise deployment inside the customer's own network is available on request for teams that need images and results to stay in-house.
What does it integrate with?
The algorithms are delivered through third-party digital pathology platforms, with Proscia Concentriq, Pathomation and Techcyte Fusion named in dedicated announcements. On the imaging side, the ZEISS Axioscan 7 workflow is supported, along with images from spatial platforms such as RareCyte Orion, Akoya PhenoImager HT and Lunaphore COMET, plus the laboratory's own LIMS or image management system.
Is there a public API?
No API documentation is published on the site, and no API is advertised as part of the offer. Integration is handled through the pathology platform vendors rather than through developer-facing endpoints.
Where is customer data hosted?
No hosting country or region is stated for customer data. The privacy policy names processors used for the website itself, but says nothing about where patient tissue images are stored, so this is worth asking before signing.
Conclusion

Should you pick Mindpeak?

Mindpeak is a serious, narrowly focused product rather than a general-purpose AI tool. It rests on a single foundation model, publishes a real body of validation work, and states the regulatory status of each algorithm in a downloadable annex, which is more candour than this market usually offers. For a laboratory already running a digital pathology platform, the integration path is genuinely short.

Three things deserve attention before committing. The first is the licence over uploaded images: it is perpetual, sub-licensable, and explicitly covers retraining the vendor's AI, with no documented way to withdraw. For a laboratory handling patient tissue, that clause belongs in front of legal counsel and the data protection officer, not in the small print. The second is the regulatory picture: most of the catalogue is Research Use Only, only nine listed products carry CE-IVD marking, and there is no clearance in the United States, Canada or the United Kingdom, so the intended use of each algorithm has to be checked one by one. The third is commercial opacity: no price, no trial, no data processing agreement and no stated hosting country, all of which have to be resolved in conversation.

The company itself is well established for its size, founded in Hamburg by Felix Faber and Dr Tobias Lang, backed by a USD 15.3 million Series A led by Zeiss Ventures, and present in the United States through a Cambridge subsidiary. Buyers who need standardised biomarker scoring at volume, and who can live with the image licence, will find a mature offer here. Buyers who want to evaluate before talking to sales will not.