The free Guest plan includes 5 monthly credits and 10 AI runs to get started. Create an account without a credit card.
A free AI run covers the full cost of one task within the model's supported limits. In a multi-file batch, one free run covers one submitted scan. In Interactive Studio, one free run covers the clicks and structures in one AI session.
Free runs are separate from credits and are used first when billing your personal account. They do not cover tasks billed to a lab. During an announced free-runs period, tasks do not use your free-run allowance or credits.
No. Run models and review results in your browser. No installation, programming, or local GPU is required.
You can refine segmentations with interactive and manual editing tools, then download results for your research workflow.
Use a compressed NIfTI file (.nii.gz) or a DICOM folder. Other research imaging formats are also supported and converted for use on the platform.
The per-file upload limit is 500 MB. Check supported formats and input requirements for the full list and the selected model's requirements.
Synthetic image generation does not require an input scan.
Follow progress in Queue, then show or download the output in Results. See the documentation for detailed steps.
If it still has not arrived, contact us and include the email address you registered with.
Processing time depends on the model, image size, and available GPU capacity. Larger scans and tasks waiting in the queue can take longer.
Check your task's progress in Queue.
Follow @nalvera_ai for model releases, new features, maintenance updates, and research notes.
Submitted AI tasks combine a fixed processing fee with a model cost based on image size. For synthetic generation, the requested output size and number of images determine the volume.
Interactive Studio clicks have no fixed processing fee; their cost depends on scan volume and the Studio rate. Drawing a structure by hand that has no AI segmentation costs the same as one click, once per structure.
Monthly credits are used first, followed by top-up credits. See Pricing for current examples.
No. Monthly credits reset on a rolling 30-day cycle, and unused credits do not carry over. This applies to the free Guest plan and paid plans.
Purchased top-up credits do not expire and remain available until used.
Yes. If a processing error stops the task, its reserved credits are returned automatically.
A completed task is not refundable because its output is unexpected or unsuitable for your study. Review the model's intended use and limitations before submitting.
Click your credit balance in the navigation sidebar, or open your account menu and choose Purchase Top-Up Credits.
Choose a preset amount or enter a custom value. The minimum purchase is €5 for paid plans and €10 for the free Guest plan.
Yes. Open Settings → Billing, then Manage in Stripe to cancel.
Your paid plan remains active until the end of the paid period. Unused time is not partially refunded.
Paid plans include discounted top-up credits. Higher plans offer a larger discount, alongside their monthly credit allowance.
Compare the current rates on Pricing.
The icon identifies a model or task linked to the Nalvera partnership program. Its developers receive an agreed fee for eligible completed runs, with no separate partnership surcharge to you.
Partnerships can apply to an entire model or individual tasks, including tasks within an open-access model. See the partner program for more information.
Unused monthly credits are forfeited when your account closes. Eligible purchased credits can be transferred to another user or lab using a one-time code, or you can choose to forfeit them.
The transfer option is available when the purchased balance exceeds the free welcome credits. Only credits older than 30 days qualify, and a 5% fee is deducted.
The code is shown once during deletion and expires after 31 days. Save it securely and share it with the recipient before it expires; unused transferred credits are then lost.
Study Design groups scans by study and patient, measures every segmented structure, follows those measurements over time and compares patient groups. Use it when your research covers more than one scan or patient.
Scan types are CT, MRI, PET, SPECT, PET-CT, SPECT-CT and Other.
A modality mismatch is flagged for confirmation. Choose a model suited to your scan and research question.
Add the scan to a study and send it to a segmentation model such as TotalSegmentator. Nalvera.AI then measures every outlined structure in your browser and lists its volume and Hounsfield values in one table, which you can copy as CSV.
Nothing. Measurements, charts, group comparisons and study reports use no credits. The AI task that makes the segmentation uses credits or a free AI run.
Yes. Attach your own mask to a scan in the study, on the same image grid. Or open the scan in the Interactive Studio, outline the structure with clicks and save. Both are measured like an AI result.
Attach the PET volume to a CT or MRI scan in a study that has a segmentation. Each structure then gets SUVmean, SUVmax and SUVpeak, plus, under Columns, SUL and other normalisations when height, weight and sex are known. A value with a missing input shows a dash. For SUV, the ! beside the structure gives the reason.
SUV is the activity concentration times body weight, divided by the injected dose corrected for decay. The scan time and the dose decay between injection and scan start follow DICOM PS3.3 and the QIBA vendor-neutral SUV pseudocode. An SUV with an unknown decay mode is not calculated; an assumed mode is flagged for you to confirm. SUVpeak is the highest mean in a 1 mL sphere.
Yes. Attach the PET or SPECT volume to the CT or MRI scan in the study and view both in the Fusion Viewer. A volume on another grid is aligned by its image headers, and you can refine that with rigid registration, which runs in your browser and is saved with the volume.
Enter each scan's date. Trends then charts a structure's volume, mean value or SUV for one patient; volume charts add the percentage change and, for growth, the doubling time. Once the scans have timepoints (Assign timepoints by date), Trends also draws one line per patient or a group mean with a standard deviation band. Each chart downloads as an image.
Add a categorical patient field such as treatment arm, enter each patient's value and choose it under Compare groups. Pick volume or another measure under Metric. Each structure measured in at least 2 patients then shows n, mean, SD, median and IQR per group, from each patient's latest value. Two groups also get a Mann-Whitney test, a false-discovery-rate q value and an effect size.
Yes, for a patient field with the role Outcome and two classes, such as responder or non-responder. Compare groups draws the ROC curve for a chosen structure and measure, with the AUC and a 2×2 table. Each class needs at least 3 patients. The result is exploratory.
The study report comes as a PDF with methods and provenance, a CSV table or a JSON file. A ZIP holds each scan's latest model result and its Studio results, with a provenance list. Each task sent to a model also records its model version for your methods section.
The Cohort tab shows a grid of patients by timepoint, with each cell marked No scan, No result, Latest run failed, Result or Quantified. Re-run failed runs again the scans that failed on the model and task you choose.
The current limit is 500 MB per upload. Larger files are rejected.
If you crop or resample a scan, check that it still meets the model's requirements and your research needs. For regularly larger datasets, contact us to discuss options.
Yes. Each file creates a separate AI task and is processed and billed separately, using a free run when eligible.
All files in a batch use the same model and settings. Use separate batches for different models.
Upload time depends on file size, your connection, and any browser-side conversion. A wired connection and pausing other uploads can help.
Follow your organisation's network requirements. If uploads repeatedly stall, contact us with the file format, approximate size, and any error message.
For segmentation, TotalSegmentator CT, NV-Segment-CT and TotalSegmentator MRI outline organs, bones and vessels across the body, MOOSE CT covers focused regions such as lung lobes and muscles, MuscleMap MRI outlines muscles and bones on whole-body MRI, MRSegmentator outlines the trunk on MRI or CT and body composition on MRI, and Nalvera iSeg segments any structure you click. MAISI-v2 generates synthetic CT or brain MRI images, and Stanford Merlin and NV-Reason-CT write research reports from CT. Compare all models.
Yes. Open Results and choose Show.
Download the output when you are ready to use it elsewhere.
A Lab adds a shared credit budget for a team. A lab manager can invite members, purchase credits, and set individual spending limits.
Each member submits tasks from their own account. Task data and results remain private to that user; the lab provides the shared billing option.
Contact us to set up a lab or discuss a Data Processing Agreement.
Lab members and lab administrators cannot access your AI tasks, input files, or segmentation outputs. Lab administrators see only your aggregate spending from the lab budget.
Lab top-ups use a shared lab rate, shown in the top-up dialog. It does not depend on the manager's personal plan.
Personal-plan top-up discounts apply only to the personal credit balance.
Your uploaded data is accessible to you and a limited number of authorised Nalvera staff under access controls and logging. Other users, including lab members, cannot access your tasks or results.
Imaging data uses EU-hosted processing and storage, with encrypted transfers to and from the website. See the Privacy Policy for access, sharing, and deletion rules.
No extended "remember me" option is available. Sessions expire after about a day, and you may need to sign in again when your browser session ends.
Always use Sign out when you finish on a shared computer.
Yes. Remove identifying information before upload and check the image itself as well as its metadata.
Nalvera clears NIfTI free-text header fields, and supported browser-side conversions discard original headers. These safeguards do not replace your own checks. See input preparation and DICOM conversion for details.
Files are normally stored for 90 days from upload. Check the deletion date in Results or Study Design. Extend storage for 1 credit per 5 extra days per dataset, using monthly, top-up or eligible lab credits.
Download files you need before expiry. File removal is separate from record retention: anonymised usage records, including the model, processing duration, and date, may remain after files are removed. Measurements taken before the deletion date stay in the study after its files are removed.
Files still referenced by another study record, or retained under a research opt-in, follow those retention rules. See the Privacy Policy for details.
Privacy Vault is in development and is not yet publicly available. It is planned as a free desktop app for preparing CT and MRI scans locally before upload.
The Privacy Vault page explains the planned pseudonymisation, optional face and embedded-text removal, and encrypted .nvlt exports.