Radiology & Imaging AI
I

iMeasure

iMeasure is Better Medicine's CT lesion measurement service, sold as an API for the viewers, PACS and trial reading platforms radiologists already use, and shown on its site under the BMVision line as BMVision XCS. A reader clicks a lesion and, in under a second, the model returns a segmentation mask with two axial diameters, height and volume in 2D or 3D, for lesions anywhere in the chest or abdomen. Interactive mode lets the reader add clicks that include or exclude tissue and redraw a boundary with MagicPen, and a two point mode serves simpler viewers.

Registration matches the same lesion across baseline and follow up scans, so a viewer can show each finding's history and compute RECIST 1.1 target lesion sums, percentage change and response. Every call returns structured data for a report, an archive or a trial database. Software makers integrate through a REST endpoint and a TypeScript SDK with OHIF examples, hospitals can use the bundled BMViewer, and sponsors and contract research organizations can engage Better Medicine directly for multicenter oncology reads. Mean 2D Dice overlap runs from 0.92 for kidney lesions to 0.78 for liver, with lymph nodes added in 3D at 0.71.

Better Medicine OÜ was founded in Tartu, Estonia, in 2020 and also sells BMVision Kidney, a CE marked kidney lesion detector. It counts Universitätsmedizin Mainz, the NHS, NIHR, Tartu University Hospital and Pärnu Hospital among the organizations it works with, along with the radiology AI marketplaces deepc and CARPL, without saying which of them run iMeasure.

AI Health Index verifiedNovember 20, 2026
Compare iMeasure with other vendors
Founded
2020
Headquarters
Tartu, Estonia
Categories
radiology-and-imaging-ai, clinical-trials-ai
Indexed Products
iMeasure 3D API, iMeasure TypeScript SDK, BMViewer RECIST 1.1 reporting
Buyer Segments
Clinical trial imaging platforms, Contract research organizations, Pharmaceutical trial sponsors, Viewer and PACS suppliers, Hospital radiology departments
Assessment

Capability Axes

An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read

AI Capability
AA on AI CentralityThe artificial intelligence is the product. Remove the model and there is nothing left to sell.
Vendor Published

Segmentation is the whole product. A viewer sends a cropped CT volume around the lesion the reader picked, and the model returns a binary mask from which the service derives the long and short axial diameters, maximum diameter, height and volume. Static mode works from two extreme points on one slice, while interactive mode carries a running set of positive and negative clicks and can send back the previous mask, or MagicPen contour strokes, to refine the result in place.

Registration then follows the finding across timepoints. The viewer around it supplies images, controls and reporting, which is why the API can drop into a PACS or trial platform. Take the model away and there is no measurement left to sell.

BB on Autonomy and Oversight ModelThe oversight structure is described and one part is missing, commonly the threshold at which the system stops or what happens after it is wrong.
Vendor Published

Nothing happens until a reader chooses a lesion. In interactive mode the reader then shapes the result with include and exclude clicks or MagicPen strokes, and the existing contour travels back with each request so corrections build on one another, while the two point mode suits viewers that want a single gesture. The model never picks lesions on its own, and longitudinal tracking matches a finding the reader has already selected.

Control loosens at reporting: BMViewer can produce a full RECIST 1.1 report automatically, and when a host platform must hold a measurement for confirmation is left to that platform. No threshold, such as a low quality segmentation, routes a contour back to a person.

BB on Model and Technology TransparencyThe approach or the suppliers are named without the version and update discipline behind them. Naming a supplier is the entry to this band both here and on Model Supply Chain Disclosure, which ask different questions of the same disclosure: who receives the data, and what produces the output.
Vendor Published

The model is promptable: a CT segmentation network takes a bounded crop of voxels plus the reader's clicks and returns a mask in the source study's coordinates, and a separate registration step matches the finding across scans. Positive and negative clicks, a prior mask and MagicPen strokes steer each prediction, and a crop smaller than 64 by 64 by 16 voxels is refused. That explains how the product works and that the model is Better Medicine's own.

Training data, the network architecture, a model version and how a changed model is announced to integrators are not given, and the API carries no version number. The accuracy figures are labeled as subject to continuous improvement, so the model does change over time.

BB on Model Supply Chain DisclosureSubstantial partial disclosure, or a chain that is structurally short: an in house build, a cleared model that cannot be quietly swapped, or a deployment where the transfer does not occur at all. Naming the model provider exits the band below into this one; the axis rises from here on the completeness of the party list and on the terms that govern data once it arrives.
Vendor Published

Better Medicine builds and hosts the segmentation model itself, so the chain from a viewer to an answer is short: the integrator's viewer sends the crop, metadata and any prior mask to Better Medicine's endpoint, and the service returns the prediction. The TypeScript SDK is Better Medicine's own, and OHIF is the open source viewer in its examples rather than a party that processes data. The hosting behind the endpoint is not named.

The cloud provider, any subprocessors and the terms that govern submitted scans once they arrive, including retention and training rights, are missing, which matters for a trial sponsor sending images from sites in several countries.

CC on Clinical and Operational EvidenceNamed customers, or vendor reported percentages with no method, denominator or reference standard. Scale of use is recorded here and is not treated as evidence of benefit.
Vendor Published

The results offered for iMeasure are segmentation overlap figures: mean 2D Dice of 0.92 for kidney lesions, 0.89 for pancreas, 0.82 for lung and 0.78 for liver, and 3D Dice from 0.85 for kidneys down to 0.71 for lymph nodes. Dice measures how closely a predicted region matches a reference outline. No cohort size, reference reader method, confidence interval or study design comes with the numbers.

The pitch rests on reader variability and the time manual RECIST measurement takes, yet no trial, site or reader study reports read time or agreement with iMeasure in use. Better Medicine's peer reviewed work at Tartu University Hospital concerns its separate kidney detector rather than this API.

CC on AI Safety and PHI StewardshipGeneral assurances of privacy and security that do not answer the questions artificial intelligence raises: what is retained, what reaches a model, and what happens to it there.
Vendor Published

Each call carries only the CT crop around the selected lesion, as raw voxels, with its geometry, scan spacing, rescale values and a request identifier, plus the current mask during refinement. The request format has no field for a patient name or identifier. Because the endpoint is stateless, every request carries its full context, and the finding history stays wherever the integrator keeps it.

The developer material promises secure handling and privacy protection for sensitive medical information without saying what that involves. How long submitted crops, masks and the annotated images stored during analysis are kept, whether they train later models, and how a safety event is handled are all left open, so the commitments stop at general assurance.

Regulatory and Compliance
DD on HIPAA and BAA PostureNo statement of status and no privacy document that reaches the product. Both halves are absent at this grade: a service level privacy document on its own lifts a vendor to the band above even where no status is stated. A website privacy notice, including one that says it does not cover customer data, does not reach the product, so a vendor whose only document is that notice and who states no status grades here.
Vendor Published

Better Medicine OÜ's privacy policy covers its website, inquiries and support correspondence, where the company acts as controller, keeping support records up to 10 years for medical device surveillance and commercial requests for 24 months after the last contact. It does not reach the CT crops and masks the API processes. The developer site says the service is built with healthcare compliance in mind and handles sensitive medical information securely, without naming a framework.

Better Medicine states no business associate status for iMeasure and offers no GDPR processor terms for it, so a US platform has no status to rely on and a European one has no processor contract to read before signing.

DD on Security Certifications and Trust CenterControls are asserted with nothing independent behind them, or nothing is published. Read the note before concluding anything: this is the grade most often corrected on a second pass, because assurance material frequently sits on a parent domain or inside an old announcement rather than on the product pages.
Vendor Published

Requests travel over HTTPS to the hosted endpoint, and the developer site calls the service secure and compliant, with a focus on data protection. The ISO 13485 mark on the product page, issued through BSI, certifies quality management for medical device development and does not audit how the service protects data. No SOC 2 report, ISO/IEC 27001 certificate, penetration test or trust center covers iMeasure, and the API's authentication scheme is not stated. The hosted service also leaves audit logging, access controls and incident notification undescribed, so nothing independent stands behind the security claims made for the endpoint.

DD on FDA and Regulatory StatusThe positioning is unclear or unsupported: clinical claims with no regulatory position stated, or language implying a status the databases do not carry. An open enforcement action that names the product also grades here, whatever the positioning: an FDA warning letter with no close out letter posted, or a consent decree in force. The note says which of the two applies, and the grade returns to the positioning bands when FDA closes the action.
Vendor Published

No regulatory position is stated for iMeasure. Its product page carries an ISO 13485 quality management logo and Better Medicine describes its portfolio as CE marked, while the developer site offers the measurements for clinical decision making and the product page sells them for lung screening, radiotherapy and surgical planning as well as trial reads. The CE marking Better Medicine holds attaches to BMVision Kidney, a separate detection product.

For the API itself there is no intended use, device class, research use limit or clearance. A PACS maker putting iMeasure diameters into a clinical report cannot tell from the product's own pages whether it is a medical device, under which regime, or for which lesions.

AA on AI Governance and Bias DisclosureA bias or fairness evaluation with a stated method, subgroup performance, or an independent audit of model behavior. Subgroup performance means published results compared across the groups the product serves, clinical and physiological as well as demographic, so a buyer can see where it underperforms.
Vendor Published

Overlap results are broken out by organ, the clinical groups a lesion measurement tool serves, so a trial imaging operator can see where the model is weaker before choosing target lesions. Mean 2D Dice runs 0.92 for kidney lesions, 0.89 for pancreas, 0.82 for lung and 0.78 for liver. The 3D results read 0.85 for kidneys, 0.82 for lungs, 0.81 for pancreas, 0.77 for liver and 0.71 for lymph nodes, which makes liver and lymph nodes the clear weak spots.

The figures carry a label saying they are subject to continuous improvement, and no cohort size, scanner mix, contrast phase or lesion size breakdown sits behind them. Patient demographics are not reported, and no outside audit of model behavior is named.

CC on AI Liability and RecourseMechanisms exist that let someone challenge an output, such as audit trails, source traceability or review before commit, with nothing standing behind the output and no route for the harmed party.
Vendor Published

The predicted mask is drawn over the original CT region in the viewer, so the radiologist sees exactly what the model included and can change it with more clicks or contour strokes before a measurement is accepted. Each response carries a request identifier and the geometry of the crop, which lets an integrator trace a measurement back to the call that produced it.

Failures come back as named error codes, such as an empty segmentation or a crop below 64 by 64 by 16 voxels, rather than as a silent result. No warranty, remedy or service credit attaches to a wrong diameter, and the API has no terms of its own outside the contract a customer negotiates, so a sponsor relying on the numbers has the reader's check and that contract.

Integration and Deployment
BB on EHR and Interoperability DepthNamed systems with read access or one directional writing, or standards support with named deployments behind it.
Vendor Published

Integration happens at the viewer. The REST endpoint accepts a cropped CT volume in DICOM patient coordinates, with orientation and spacing, and returns a mask, geometry and measurements, while a TypeScript SDK maps those into a viewer's own segmentation objects. OHIF, the open source web viewer, is the named example in the SDK material.

Structured output from every call can feed a reporting system, archive or trial database, and BMViewer carries a full RECIST 1.1 workflow for buyers who want it bundled. No named PACS vendor, trial imaging platform or hospital is shown running the API in production, and no clinical record system connection is part of the product.

CC on Deployment Model and Data ResidencyA single hosted option with location implied rather than committed.
Vendor Published

Inference runs on a hosted Better Medicine endpoint, which receives the cropped image region from the viewer and returns the mask and measurements. Since requests are stateless, the integrator decides where the finding history and masks live on its own side, and BMViewer serves buyers without a viewer of their own. The country where inference runs, the cloud behind the endpoint, where annotated images rest and how one customer's requests are kept apart from another's are not stated.

No on premise or private cloud option is offered. Better Medicine is based in Estonia, and that location is the only geographic fact attached to a service that trial sponsors would feed with scans from many countries.

Commercial
CC on Commercial TransparencyNo price is published and the posture is discoverable: a buyer can establish how the product is sold and what drives the cost before contacting the vendor. Most of the index sits here.
Vendor Published

Three ways to buy are on offer: the API for a software maker's own viewer or PACS, BMViewer as an end to end option with RECIST 1.1 reporting, and direct support for oncology trials across sites and readers. Each starts with a demo or a conversation with Better Medicine. How the API is charged, whether per call, per study, per site or per seat, is not given, nor is a minimum commitment, a separate price for BMViewer or a fee for trial services.

The audience is clear enough to scope that conversation: viewer and PACS component makers, lung screening and planning software, hospitals, contract research organizations and pharmaceutical sponsors, each matched to one of the three routes.

BB on Setting and Specialty CoverageCoverage is named with validation behind part of it.
Vendor Published

CT is the only modality, for solid lesions in the chest and abdomen, with organ results for kidney, pancreas, lung, liver and lymph nodes. The settings named are oncology response assessment under RECIST 1.1 in trials and routine follow up, lung cancer screening platforms, and organ and lesion segmentation for radiotherapy and surgical planning. The organ overlap figures support part of that range.

Nothing supports performance in screening or planning workflows specifically, no MRI or PET path exists, and lesions in other chest and abdominal organs rest on the claim that the model handles any of them. Buyers are viewer makers, trial platforms, hospitals, contract research organizations and pharmaceutical sponsors.

Alternatives

Alternatives to iMeasure

Other products that do the same kind of work as iMeasure, and what each one does.

Better Medicine builds AI for reading abdominal CT, starting with BMVision Kidney, which detects kidney lesions as small as 5 mm, draws their boundaries in 3D, measures the longest dimension and three orthogonal axes, and separates likely malignant from benign findings for a radiologist to review.

CARPL is a vendor neutral platform that lets radiology departments and imaging groups buy, test, deploy and monitor other companies' imaging AI through one integration and one viewer.

Beijing based imaging AI company whose InferRead line detects findings on CT and radiographs (lung nodules on chest CT, chest abnormalities on radiographs, coronary stenosis on CTA, hemorrhage on head CT, chest fractures and pneumonia), alongside the InferOperate suite of 3D…

AISAP sells AI software for cardiac ultrasound at the point of care, built at the Heart Center of Sheba Medical Center in Ramat Gan, Israel.

MosaicOS is a cloud radiology platform from Mosaic Clinical Technologies, the technology subsidiary Radiology Partners launched in July 2025.

New York company, founded in 2024, building deep learning tools for PSMA PET/CT in prostate cancer and theranostics.

Commercial

Pricing

Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.

Entry Price Pricing Basis BAA Tier Implementation Source
Contact sales
iMeasure is sold three ways: as an API for an existing viewer, PACS or trial reading platform, inside the BMViewer viewer with RECIST 1.1 reporting, or as trial support for sponsors and contract research organizations. Prices are quoted after a demo, and the unit of charge is not given. No business associate agreement or GDPR processor terms are offered for iMeasure. Better Medicine's privacy policy covers its website, inquiries and support correspondence, not the CT images the API processes. Integration runs through a REST endpoint and a TypeScript SDK with OHIF examples, and BMViewer suits buyers without a viewer of their own. No setup, integration or onboarding fee is stated. Vendor Published

No per call, per study, per site or per seat price is shown, and no minimum volume, contract length or free trial is offered. Better Medicine sells the API and BMViewer through demos and direct conversations.