Aiosyn Mitosis Breast
Aiosyn Mitosis Breast is a deep-learning model that finds mitotic figures across an entire breast H&E whole-slide image and marks them before the pathologist opens the case. The product page labels it research use only.
What is Aiosyn Mitosis Breast?
Aiosyn Mitosis Breast is a single deep-learning model rather than a platform: it detects mitotic figures in breast tissue sections stained with haematoxylin and eosin. Mitotic count is one component of Nottingham grading and a prognostic indicator in breast cancer, yet counting by hand is slow, samples only a handful of fields, and varies from one reader to the next. The model inverts that. It runs over the whole-slide image, every field, and returns each figure it finds already marked on the slide before the case is opened, so the pathologist confirms a count instead of hunting for candidates.
It ships as modular software offered as a service. Nothing changes upstream: sections are prepared and scanned exactly as they already are, with no second scanning workflow and no new file format to produce. Read-outs come back in the viewer the laboratory already runs, and the analysis executes either inside the laboratory's own network or on the vendor's platform, whichever its governance requires. The named delivery routes are Sectra Amplifier Marketplace, Telemis TM-Microscopy, Techcyte Fusion, and inline execution on Pramana scanners whose edge computing returns results as DICOM annotations.
The model was assessed in a multi-centre clinical performance study run at Radboud university medical center: 28 pathologists from nine countries read 210 whole-slide images drawn from eight international centres, each set twice, with and without the algorithm, separated by a three-week washout. Where the numbers come from matters here. The product page and the vendor's press releases report around 60% time savings, 15% higher productivity and 32% better inter-observer consistency, with more than 90% of surveyed participants willing to adopt it. The scientific abstract published on the same site reports a 10.4% time gain, a 15.5% productivity increase on resections, a significant fall in inter-observer variability and no loss of accuracy. All of it originates with the vendor and with Radboudumc, the centre Aiosyn was spun out of.
Regulatory status is the second thing to read carefully. The product page and the Clinical page both state that the model is for research use only, while a January 2025 announcement on the same site claims CE marking under the IVDR for diagnostic use. Aiosyn publishes several other pathology models; this record covers only this one.
What it does
- Detect mitotic figures in breast tissue sections stained with haematoxylin and eosin
- Read the entire whole-slide image rather than a sampled field
- Mark every detected figure directly on the slide
- Deliver the detections before the pathologist opens the case
- Return read-outs inside the viewer the laboratory already uses
- Run in the cloud or on premises, inside the laboratory's own network
- Run inline on the scanner and hand back DICOM annotations, in the Pramana and Techcyte Fusion configuration
When to use Aiosyn Mitosis Breast / When not to
A quick filter to help you decide if Aiosyn Mitosis Breast is the right fit.
When to use Aiosyn Mitosis Breast
- Pathologists grading breast carcinoma, for whom mitotic count is a required, slow and poorly reproducible part of Nottingham scoring
- Digitised pathology laboratories already running Sectra, Telemis or Techcyte Fusion, where the model drops into the viewer they use every day
- Research pathology groups, which is the use the product page explicitly designates the model for
- Hospital networks evaluating AI under real clinical conditions, as ZAS Netwerk did across three of its breast clinics
- Laboratory managers facing rising workloads, longer turnaround times and the global shortage of pathologists
When not to use Aiosyn Mitosis Breast
- Laboratories that are not digitised: the model needs whole-slide images from a scanner, not glass under a microscope
- Anyone working outside breast tissue or outside H&E staining, which the vendor redirects to a bespoke development engagement
- Buyers who need a published price: no figure, no tier and no pricing page exists anywhere on the site
- Teams wanting a public API to wire up themselves, since access runs exclusively through partner platforms
- Anyone looking for slide quality control or kidney analysis, which are separate products in the same catalogue
How to use Aiosyn Mitosis Breast
A typical end-to-end flow, from setup to results.
- Read the dedicated mitotic figure counting page on the Aiosyn site to check the model matches your tissue and stain
- Request a demonstration through the booking form, choosing "Mitosis in breast cancer" as the area of interest
- Send a handful of your own whole-slide images; Aiosyn runs the analysis and shows the detections in its viewer alongside your images, with no technical work on your side
- Check whether the platform you already run carries the model: Sectra Amplifier Marketplace, Telemis TM-Microscopy or Techcyte Fusion
- Settle the deployment mode with your IT and governance teams, cloud or an on-premise installation inside your own network
- Set up the integration; once it is in place, data transfer runs automatically
- Carry on preparing and scanning sections as before, with no second scanning workflow and no new file format
- Let the algorithm detect mitotic figures across every field of the whole-slide image
- Open the case and find each detected figure already marked on the slide
- Confirm the count: the pathologist remains the decision-maker
Pros & Cons
Pros
- A published multi-centre validation study, 28 pathologists from nine countries reading 210 slides from eight centres, with abstract, authors and poster on the site rather than a bare marketing claim
- Detection across the entire slide, where manual counting samples only a few fields
- Measured improvement on inter-observer consistency, the acknowledged weak point of mitotic scoring
- No loss of accuracy in the published abstract, reported as non-inferiority at p<0.001
- Fits the existing workflow: no second scan, no new file format, no change to how slides are prepared
- Four named delivery routes (Sectra, Telemis, Techcyte, Pramana) rather than a single proprietary viewer
- On-premise deployment available for laboratories unwilling to send images outside their network
Cons
- The site contradicts itself on regulatory status: research use only on the live product page, CE-marked under the IVDR in three separate 2025 announcements
- The 60% time saving on the product page is not the 10.4% reported in the abstract published a few clicks away on the same site
- Every performance figure comes from the vendor and from Radboudumc, the centre it was spun out of, not from an evaluation independent of the vendor
- No price is published at all: no grid, no order of magnitude, no entry point
- No terms and conditions and no legal notice, so ownership of slide images and of the detections is addressed nowhere publicly
- No hosting country or region is declared for customer data, and nothing is said about whether customer data is used to train the models
- Narrow scope and a heavy prerequisite: breast H&E only, and only for laboratories that are already digitised
Pricing & Plans
No price is published. The site carries no pricing page, an absence confirmed against its complete sitemap, and no amount, currency or tier appears anywhere on it. Neither a permanent free plan nor a free trial is announced. The only free access documented is a demonstration in which Aiosyn runs the analysis on whole-slide images the laboratory sends. A price is obtained by requesting a demonstration, contacting the vendor directly, or going through the marketplaces of the partner platforms.
Data, GDPR & hosting
A consolidated view of how Aiosyn Mitosis Breast handles your data.
GDPR overview
Implementation is explicit and reasonably detailed. The privacy policy, last amended on 24 August 2026, cites Regulation (EU) 2016/679 by name, lists four legal bases (consent, performance of an agreement, legal duty, legitimate interest) and sets out eight data-subject rights, including erasure, portability and the right to complain to the Dutch supervisory authority, the Autoriteit Persoonsgegevens. Aiosyn B.V. is identified as controller with its Chamber of Commerce number and a named contact person. Cloudflare, Inc. and Jira Service Management are named as processors, with transfers outside the European Economic Area covered by the European Commission's standard contractual clauses. The site sets no cookies of its own and shows no consent banner. One limit deserves stating plainly: the policy covers the website and preclinical research services, not the handling of slide images by the product itself.
Who owns the data?
The site publishes no terms and conditions, no legal notice and no framework agreement, an absence confirmed against its complete sitemap rather than merely unobserved. Nothing therefore states who owns the slide images sent to Aiosyn, who owns the detections the model returns, or what either party may do with them: those questions are settled in a contract negotiated offline. The only published legal document is the privacy policy, and its scope is the website and preclinical research services. It names Aiosyn B.V., registered with the Dutch Chamber of Commerce under number 83481435, as controller, states that personal data is processed only by people working for and on behalf of Aiosyn, and is governed by Dutch law.
Reuse rights
No reuse clause is published, because no terms of use are published at all. What a laboratory may do with the detections, and what Aiosyn may do with the images it receives, is addressed nowhere on the site; it belongs to the contract that follows a demonstration. Two published facts bear on the question. First, the analysis can be deployed on premises, in which case, in the vendor's own words, it runs inside your own network rather than on the vendor's platform, and the choice is presented as the customer's governance decision. Second, results are returned into the laboratory's existing viewer or into the partner's product, and in the Techcyte and Pramana configuration they arrive as DICOM annotations, a durable and portable form. For personal data collected through the website, the privacy policy grants the full set of GDPR rights, erasure and portability included.
Data retention & training
Hosting summary
No hosting country or region is declared for customer data. The vendor makes the location a customer decision: the analysis runs either inside the laboratory's own network or on Aiosyn's platform, whichever its governance requires, and in the Techcyte and Pramana configuration it runs inline on the scanner itself, returning DICOM annotations. Nothing more precise is published about where slide images are processed or stored. For the website alone the picture is documented. Cloudflare, Inc. hosts and delivers the site and processes the full IP address of every request; transfers of personal data outside the European Economic Area are covered by the European Commission's standard contractual clauses. Amazon Web Services, Microsoft Azure and CloudNation appear as partner logos on the partnerships page, which is not a statement about where data is hosted, and a host's certifications are not the vendor's own. Aiosyn B.V. is established in Nijmegen, in the Netherlands. Its privacy policy is governed by Dutch law and names the Dutch supervisory authority as the competent regulator.
Things to keep in mind
Risks and trade-offs to weigh before adopting Aiosyn Mitosis Breast.
- Mitotic count feeds Nottingham grading, which feeds prognosis and treatment choice, so an error here is not a cosmetic one
- Regulatory ambiguity is the practical risk: using a model the vendor's own product page labels research use only to support a diagnosis puts the burden squarely on the laboratory
- Automation bias, because reviewing pre-marked candidates is not the same act as searching, and a reader can drift into validating whatever the machine proposes
- The mirror risk is the worse one: mitoses the model does not mark stand less chance of being noticed by a reader who trusts the pre-marking
- Skill erosion, since a task performed less often is performed less well, including on the cases where the model is unavailable
- The performance figures come from the vendor and its parent centre and are presented differently on different pages of the same site, so they are not independent measurements
- Data governance, because slide images can be identifying, no terms of use are published, no hosting country is declared, and nothing is said about whether customer data trains the models
Setup & Integrations
Technical difficulty
Low for a trial, moderate for production. A demonstration takes no technical work at all: the laboratory sends whole-slide images and Aiosyn runs the analysis. Going live is an integration project, with Sectra, Telemis or Techcyte, or an on-premise installation, plus a governance decision on whether images leave the building. Once the integration is in place, data transfer is automatic and nothing changes upstream. The real hurdle is the prerequisite: the laboratory must already be digitised, with a scanner and an image management system. There is no self-service sign-up.
Deployment
Integrations
Behind Aiosyn Mitosis Breast
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What exactly does Aiosyn Mitosis Breast detect?
Does it replace the pathologist?
Can it be used for diagnosis?
What improvements were actually measured?
Who ran the validation study?
Which software does it work with?
Can it run inside our own network?
What does it cost?
Does it work on tissues other than breast?
Is there an API?
Should you pick Aiosyn Mitosis Breast?
Aiosyn Mitosis Breast is unusually narrow and unusually well documented. One tissue, one stain, one task, and behind it a multi-centre study whose abstract, authors and poster the vendor publishes rather than merely summarising. For a pathology laboratory already digitised and already running Sectra, Telemis or Techcyte Fusion, the case is easy to follow: exhaustive detection across the slide instead of a sampled field, measurably less disagreement between readers, and no change to how sections are prepared or scanned.
Two things deserve a direct question to the vendor before anything else. The first is regulatory status. The live product page says research use only, while three announcements from 2025 on the same site describe the model as CE-marked under the IVDR for diagnostic use. No certificate number and no notified body appears anywhere, so the claim cannot be checked from the site alone. The second is the headline figure: 60% time savings on the product page against 10.4% in the abstract published a few clicks away. The productivity and consistency numbers line up well between the two; the time saving does not.
Nothing about pricing is public, and the site publishes no terms and conditions, no legal notice and no statement of who owns the slide images or the detections. All of that is negotiated offline, starting from a demonstration the vendor runs on your own slides.
Read as what it is, a focused and credibly validated detection model from a Radboudumc spin-off sold business to business, it is a serious proposition. Read as a finished diagnostic device on the strength of the press releases alone, it is not, and the vendor's own product page says so.
- Choosing a selection results in a full page refresh.
- Opens in a new window.