About Nitsor
We make the software where experts review model suggestions on CT and MRI scans, and where training datasets are fixed from that reviewed work.
Real scans: Gruber et al., Fraunhofer EZRT (CC BY 4.0 (opens in a new tab)), Armato et al., "Data From LIDC-IDRI", TCIA (CC BY 3.0 (opens in a new tab)), Schut et al., CWI and GREEFA (CC BY 4.0 (opens in a new tab)), modified. White: published labels; dashed: model proposals from SAM 2.1, outside Nitsor. Sources
Why Nitsor exists
The system of evidence for AI-assisted inspection.
Nitsor connects CT and MRI labelling, expert review and the history behind each training dataset.
- Inspection
- An expert examines a scan and marks the structures or defects relevant to their work.
- AI-assisted
- A model suggests a label. An expert decides whether to accept, correct or reject it.
- Evidence
- The source scan, label revision, model proposal and review decision behind a training dataset.
Product decisions
What we choose to keep together.
Choose the storage that fits your work.
Reference existing scans where they live. Upload to your storage or a Nitsor-managed bucket, or let Nitsor manage all storage. Hosted uploads use Nitsor storage by default, which the owner can switch off. The browser uses short-lived signed links, and model jobs use temporary copies.
Reviews stay with the revision checked.
Submitting labels saves a revision with its author, its time and the revision before it. Each verdict on a label records the version of the label it judged. A dataset version ties those decisions to the labels used for training.
A model suggests; a person decides.
A confidence score is not a measured error rate, so every suggestion is recorded and only a person's acceptance moves it into the accepted set.
The record for three public scans: an airframe section, a chest CT and an apple. Each scan is a commit. A model's proposal, drawn dashed (SAM 2.1, run outside Nitsor), sits on a branch of its own and is never merged. Each published label is its own branch: the four radiologists who outlined the chest CT keep four separate branches, and the apple, whose dataset publishes a browning score rather than an outline, carries the score. The labels join the dataset line, a tag pins that version, and a training job reads the tag.
Hashes are sha256 of the files shown.
Our team
The people behind Nitsor.
Nitsor is led by Alberto García.
Tell us which scan workflow is slowing your team down. We can work through the labelling, review and integration requirements with you.
Terms
Choose how you run Nitsor.
Run Community on your own infrastructure for free, choose hosted Team for $649 per workspace per month or $6,228 upfront per workspace per year ($519 per month equivalent), with 250 model-assisted volumes per monthly billing cycle, or Enterprise from $12,000 per year, plus your volume commitment. No plan charges per person. What we will never meter or take away lives on one dated page, our hosting and pricing commitments. These are launch prices. Paid checkout and usage charging are not active.
Nitsor Community is free to self-host under its source-available licence.
- Read the FAQStorage, formats, review and pricing.
- Nitsor journalHow Nitsor works, with examples on public scans.
Real scans with their published labels: Gruber et al., Fraunhofer EZRT (CC BY 4.0 (opens in a new tab)), Armato et al., "Data From LIDC-IDRI", TCIA (CC BY 3.0 (opens in a new tab)), Fedorov et al., QIN-PROSTATE-Repeatability, TCIA (CC BY 4.0 (opens in a new tab)), modified.
Built and tested on public scans.
Our examples name their public datasets and sources. Where we report a measurement, the article explains the method.
Sources
Sources and licences.
These are the sources and licences for the scans and pictures on this site. Measurement articles identify the files and methods used.
Wherever a picture on this site draws a model's output, it is dashed and comes from one of the models listed below, run by Nitsor outside the product on these public scans. Labels a dataset's annotators published are drawn solid white.
Product examples on this site use placeholder people, project names, times and version IDs, and the review histories built on them are worked examples. The scans, published annotations and model runs are real, and each one is credited where it appears.
| Dataset | What it is | Scale | Where used | Credit, source and licence |
|---|---|---|---|---|
| BUS-BRA | Real 2D breast ultrasound PNG images, each with the lesion outline the ultrasonographer drew. It is not a Nitsor import or editor recording. | One 2D breast ultrasound image and its lesion mask, 385 × 393 pixels | /solutions/ | Gómez-Flores, Gregorio-Calas and Pereira. doi:10.5281/zenodo.8231412 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: image bus_0852-s enlarged 2x without interpolation, its published lesion mask outlined in white, and a SAM 2.1 proposal from one box, run by Nitsor outside the product, drawn dashed. The dataset copyright and permission notice accompanies the image. |
| JNU-IFM, video 20200103T104919 | Frames sampled from intrapartum transperineal ultrasound videos, with published pubic symphysis and fetal head masks on every frame. Nitsor does not import video. | 24 frames sampled from one video (every 10th frame), 1295 × 1026 pixels | /solutions/ | Lu, Bai, Zhou and Zhou. doi:10.6084/m9.figshare.14371652 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: cropped, the published masks outlined in white, and a SAM 2.1 proposal from one box on the first frame, run by Nitsor outside the product, drawn dashed. |
| B-mode-and-CEUS-Liver, LiverUS-108 | Liver ultrasound from a public research collection: uncompressed single-frame DICOM with Ultrasound Region calibration. It carries no labels, so none is drawn. | One single-frame image, 960 × 720 pixels; 0.2284 mm per pixel from its Ultrasound Region | /, /solutions/ | Eisenbrey, Lyshchik and Wessner, The Cancer Imaging Archive. doi:10.7937/TCIA.2021.v4z7-tc39 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: cropped, leaving out the scanner's greyscale bar, depth scale and focal-zone marker (16:9 on the home page), the scanner model name blacked out. |
| Fraunhofer EZRT XXL-CT, Me 163 subvolume V5, with published instance masks | Industrial CT of a historic airframe section, with published instance masks for 169 parts. | 512 slices of 512 by 512; 0.33 mm in plane, 0.6 mm apart | Every page (header menu), /, /about, /blog, /blog/ | Gruber, Reims, Hempfer, Gerth, Böhnel, Fuchs, Salamon and Wittenberg, Fraunhofer EZRT. doi:10.5281/zenodo.10651746 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: windowed and cropped, outlines traced from the instance masks, surfaces computed from the scan. |
| LIDC-IDRI, cases LIDC-IDRI-0002 and LIDC-IDRI-0003 | Chest CT from a public research collection. LIDC-IDRI-0003 carries four radiologists' outlines of one nodule and their ratings. | LIDC-IDRI-0003: 140 slices of 512 by 512; 0.82 mm in plane, 2.5 mm apart | /, /about, /blog, /blog/ | Armato et al., "Data From LIDC-IDRI", The Cancer Imaging Archive. doi:10.7937/K9/TCIA.2015.LO9QL9SX (opens in a new tab), CC BY 3.0 (opens in a new tab), modified: lung-windowed and cropped, reader outlines redrawn from the XML, agreement maps and boxes derived from them. |
| QIN-PROSTATE-Repeatability, cases PCAMPMRI-00001 and PCAMPMRI-00006 | A repeat-scan prostate MRI study with published manual segmentations of the gland, its zones and lesions. | 30 T2 axial slices per study; 0.2734 mm in plane, 3 mm apart | Every page (header menu), /about, /solutions/ | Fedorov et al., The Cancer Imaging Archive. doi:10.7937/K9/TCIA.2018.MR1CKGND (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: slices windowed and cropped, the published segmentations drawn as outlines. |
| CT scans, slice photographs and browning scores of 120 'Kanzi' apples, apple 98 | Food X-ray CT of apples, each with a browning score a person gave after slicing it. The dataset publishes no outline, so none is drawn white. | 629 slices of 682 by 672; 0.129 mm voxels | /, /about, /pricing, /blog, /blog/ | Schut, Trull and Couvée, CWI and GREEFA. doi:10.5281/zenodo.8167285 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: windowed and cropped, photograph resized, surface computed from the scan. |
| SPIDER: lumbar spine segmentation in MR images | Lumbar spine MRI with published vertebra and disc masks and radiologists' degeneration grades. | 447 lumbar MRI series | /, /solutions/ | van der Graaf et al., Radboud University Medical Center. doi:10.5281/zenodo.10159290 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: slices resampled to square pixels, windowed and cropped, masks outlined. |
| Data from: Deep learning unlocks X-ray microtomography segmentation of multiclass microdamage in heterogeneous materials | Composite laminates under load, with damage in three ply classes traced by trained annotators. | Synchrotron CT of composite laminae across load steps | /, /solutions/ | Kopp, Joseph, Ni, Roy and Wardle, MIT. doi:10.5061/dryad.ffbg79cwb (opens in a new tab), CC0 (opens in a new tab), modified: slices cropped and dimmed, or damage labels projected through the thickness; damage labels drawn in white. |
| X-ray CT scans of citrus fruits and their individual tissues | Citrus CT with the authors' tissue and oil-gland segmentation, made by their thresholding and morphology scripts with settings chosen by hand for each fruit and checked by eye (method: Amézquita et al., doi:10.1002/ppp3.10333). No person drew those masks, so they are drawn dashed. | CT of whole citrus fruits and their separated tissues | /, /solutions/ | Amézquita, Quigley, Ophelders, Seymour, Munch and Chitwood, Michigan State University. doi:10.5061/dryad.34tmpg4n6 (opens in a new tab), CC0 (opens in a new tab), modified: slices rebuilt from the tissue files and windowed, the authors' scripted masks outlined dashed. |
| Human Bony Labyrinth: co-registered CT and micro-CT, surface models and anatomical landmarks | Temporal bone micro-CT with an expert labyrinth label and anatomical landmarks placed by an otologist. | Micro-CT at 0.061 mm | /, /solutions/ | Wimmer, Anschuetz, Weder, Wagner, Delingette and Caversaccio, Bern University Hospital, University of Bern and Inria. doi:10.5281/zenodo.3355272 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: resliced and windowed, the expert label outlined, a surface rendered from the expert label, landmarks marked. |
| Adults' dental cone beam CT images for detecting and classifying missing teeth (V2) | Dental cone-beam CT with experts' 3D boxes around missing-tooth regions. | 85 annotated cone-beam CT scans | /workspace | Feng, Li, Gu, Wang and Yu, Science Data Bank. doi:10.57760/sciencedb.26465 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: an axial maximum-intensity slab, cropped, the expert boxes drawn as their footprints. |
| HFValid collection: hip-fracture validation collection (v3) | Femur CT with published anatomical landmarks. | 101 femur CT | /workspace | Aldieri, La Mattina, Szyszko, Baruffaldi and Viceconti, University of Bologna and IRCCS Istituto Ortopedico Rizzoli. doi:10.6092/unibo/amsacta/8431 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: a coronal maximum-intensity projection, the published landmarks drawn as crosses. |
| Wood micro-CT, Koddenberg 2025, Populus | Micro-CT of poplar wood: vessels and fibre cell walls, with the owner's tissue mask. | 500 slices of 500 by 500; 1 µm voxels | Every page (header menu) | Tim Koddenberg, University of Göttingen. doi:10.5281/zenodo.16525101 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: windowed, cropped and drawn on by Nitsor. |
| Walnut cone-beam CT, Walnut1 | Cone-beam CT of walnuts, a collection designed for machine learning. | 501 reconstructed slices of 501 by 501 | Not on a current page | Der Sarkissian, Lucka, van Eijnatten, Colacicco, Coban and Batenburg, CWI. doi:10.5281/zenodo.2686726 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: windowed, cropped and drawn on by Nitsor. |
| Scots pine seed MRI | High-field zero-echo-time MRI of a Scots pine seed. | 23.4 µm isotropic voxels | Not on a current page | Tuomainen, Himanen, Helenius, Kettunen and Nissi, University of Eastern Finland. doi:10.5281/zenodo.5541369 (opens in a new tab), CC BY 4.0 (opens in a new tab), modified: windowed, cropped and drawn on by Nitsor. |
| AOMIC-ID1000 on OpenNeuro, sub-0001 T1-weighted | Brain MRI of a volunteer from an open research collection. | One brain MRI volume, defaced by its owners | Not on a current page | Snoek, van der Miesen, van der Leij, Beemsterboer, Eigenhuis and Scholte, OpenNeuro. doi:10.18112/openneuro.ds003097.v1.2.1 (opens in a new tab), CC0 (opens in a new tab), modified: windowed, cropped and drawn on by Nitsor. |
Data citations
The Cancer Imaging Archive asks for each collection's own citation, the BUS-BRA notice asks for its article, and JNU-IFM is cited by its data article.
Wilfrido Gómez-Flores, Maria Julia Gregorio-Calas, and Wagner Coelho de Albuquerque Pereira, "BUS-BRA: A Breast Ultrasound Dataset for Assessing Computer-aided Diagnosis Systems", Medical Physics, 2023. DOI: 10.1002/mp.16812
Lu Y, Zhou M, Zhi D, Zhou M, Jiang X, Qiu R, Ou Z, Wang H, Qiu D, Zhong M, Lu X, Chen G, Bai J. The JNU-IFM dataset for segmenting pubic symphysis-fetal head. Data in Brief 41 (2022) 107904. DOI: 10.1016/j.dib.2022.107904
Eisenbrey, J., Lyshchik, A., & Wessner, C. (2021). Ultrasound data of a variety of liver masses [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/TCIA.2021.v4z7-tc39
TCIA: Clark K, Vendt B, Smith K, Freymann J, Kirby J, Koppel P, Moore S, Phillips S, Maffitt D, Pringle M, Tarbox L, Prior F. The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository. Journal of Digital Imaging 26(6) (2013) 1045-1057. DOI: 10.1007/s10278-013-9622-7
Armato III, S. G., McLennan, G., Bidaut, L., McNitt-Gray, M. F., Meyer, C. R., Reeves, A. P., Zhao, B., Aberle, D. R., Henschke, C. I., Hoffman, E. A., Kazerooni, E. A., MacMahon, H., Van Beek, E. J. R., Yankelevitz, D., Biancardi, A. M., Bland, P. H., Brown, M. S., Engelmann, R. M., Laderach, G. E., Max, D., Pais, R. C. , Qing, D. P. Y. , Roberts, R. Y., Smith, A. R., Starkey, A., Batra, P., Caligiuri, P., Farooqi, A., Gladish, G. W., Jude, C. M., Munden, R. F., Petkovska, I., Quint, L. E., Schwartz, L. H., Sundaram, B., Dodd, L. E., Fenimore, C., Gur, D., Petrick, N., Freymann, J., Kirby, J., Hughes, B., Casteele, A. V., Gupte, S., Sallam, M., Heath, M. D., Kuhn, M. H., Dharaiya, E., Burns, R., Fryd, D. S., Salganicoff, M., Anand, V., Shreter, U., Vastagh, S., Croft, B. Y., Clarke, L. P. (2015). Data From LIDC-IDRI [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/K9/TCIA.2015.LO9QL9SX
The authors acknowledge the National Cancer Institute and the Foundation for the National Institutes of Health, and their critical role in the creation of the free publicly available LIDC/IDRI Database used in this study.
Fedorov, A; Schwier, M; Clunie, D; Herz, C; Pieper, S; Kikinis, R; Tempany, C; Fennessy, F. (2018). Data From QIN-PROSTATE-Repeatability. The Cancer Imaging Archive. DOI: 10.7937/K9/TCIA.2018.MR1CKGND
Collection and sharing of these data were supported by award U01 CA151261, PI Fiona Fennessy.
| Picture | What it is | Where used | Credit, source and licence |
|---|---|---|---|
| Nondestructive Inspection of the Future (Computed Tomography), DVIDS 715683 | Footage of a portable industrial X-ray tube in an inspection shop, 2019. | / (hero film) | Tom Jozwiak, 375th Air Mobility Wing, US Air Force, via DVIDS. Source (opens in a new tab), Public domain (DVIDS) (opens in a new tab), modified: graded and cropped; maker plates blurred. |
| JSC Materials Evaluation Laboratory Capabilities (jsc2021m000789) | Footage of a materials evaluation laboratory. | / (hero film), /solutions/ | NASA Johnson Space Center. Source (opens in a new tab), Public domain (NASA) (opens in a new tab), modified: single frames cropped and graded. |
| Fresh Peaches on Conveyor Belt in Processing Plant (Pexels 32625693) | Footage of peaches under spray on a packing-line washer. | /contact | Jonathan David, Pexels. Source (opens in a new tab), Pexels License (opens in a new tab), modified: graded and cropped. |
| Colorful Peaches on Conveyor in Processing Plant (Pexels 32642674) | Footage of peaches on a grading line. | / (hero film) | Jonathan David, Pexels. Source (opens in a new tab), Pexels License (opens in a new tab), modified: graded, cropped and slowed to 0.4× source speed. |
| Man using Computer (Pexels 6236749) | Footage of a person reading CT images at a workstation. | / (hero film) | Tima Miroshnichenko, Pexels. Source (opens in a new tab), Pexels License (opens in a new tab), modified: cropped, graded and slowed; private and brand details blurred. |
| Empty MRI scanner (Pexels 7088517) | Footage of an empty MRI scanner bore. | / (hero film), /solutions/ | Pexels contributor. Source (opens in a new tab), Pexels License (opens in a new tab), modified: graded, cropped and slowed; marks blurred. |
| Empty MRI scanner (Pexels 7089596) | Footage of an MRI room. | / (hero film) | Pexels contributor. Source (opens in a new tab), Pexels License (opens in a new tab), modified: graded and slowed; marks blurred. |
| Diamond by Peter Smyth | A synchrotron experimental hall. The facility does not endorse Nitsor. | /solutions/ | Peter Smyth, via Wikimedia Commons. Source (opens in a new tab), CC0 (opens in a new tab), modified: graded and cropped. |
| Fluorescent penetrant inspection, 31st Aircraft Maintenance Squadron, 2013 | Fluorescent penetrant inspection of an aircraft part under UV light. | Not on a current page | U.S. Air Force photo by Airman 1st Class Matthew Lotz. Source (opens in a new tab), Public domain (US Air Force) (opens in a new tab), modified: graded and cropped. |
| A row of turbine blades | A row of turbine blades. | Not on a current page | Jakob Pfalz, Unsplash. Source (opens in a new tab), Unsplash License (opens in a new tab), modified: graded and cropped. |
| A clinical scanner bore | A clinical scanner bore, seen from behind. | Not on a current page | Jan Canty, Unsplash. Source (opens in a new tab), Unsplash License (opens in a new tab), modified: graded and cropped. |
| Model | What it is | Where used | Credit, source and licence |
|---|---|---|---|
| SAM 2.1, hiera-tiny checkpoint | Outlines proposed from one box, and the features its image encoder computes. | /, /about, /compare, /solutions/ | Meta AI (Ravi et al.). Source (opens in a new tab), Apache 2.0 (opens in a new tab). |
| TotalSegmentator v2.18, task total | Per-voxel anatomical classes and their softmax on the LIDC-IDRI-0003 chest CT. | /, /versioning, /blog/ | Wasserthal et al., University Hospital Basel. Source (opens in a new tab), Apache 2.0 (opens in a new tab). |
Show us what you inspect.
Start with a public CT or MRI scan. See how a model suggestion becomes a reviewed label and a fixed training dataset.