bio.rodeo
ModelsOrganizationsProvidersLeaderboardAboutSign in
bio.rodeo

The authoritative source for evaluating biological foundation models. No hype, just honest analysis.

Categories
  • DNA & Gene
  • RNA
  • Protein
  • Small molecule
  • Single-cell
  • Spatial omics
  • Pathology
  • Imaging
  • Metabolomics
  • Biosignals
  • Language model
bio.rodeoModelsOrganizationsProvidersLeaderboardAboutFAQSubmit a modelTermsPrivacyContact
© 2026 Pulsatance. All rights reserved. ~
Built by Pulsatance
models / imaging / screener
ImagingPathology
Mohamed bin Zayed University of Artificial IntelligenceReleased February 2025

Screener

Self-supervised CT foundation model that segments pathology as visual anomalies, pretrained on 30,000 unlabeled 3D volumes with no annotation.

The short version

  • —Flags pathologies nobody labeled, instead of only the handful a training set annotates
  • —Reaches 0.96 voxel-level AUROC on lung nodules with no labels used anywhere
  • —Learned masking-invariant features replace hand-crafted positional encodings as conditioning
  • —A normalizing flow scores how unlikely each voxel's dense descriptor is
  • —Fine-tunes into a segmenter that works from 25 labeled cases per fold
61Openness

Where to run it

No providers recorded yet. Browse all providers

A supervised segmentation model can only find what its training set labels. In 3D medical imaging that is a hard ceiling: annotated CT datasets cover a handful of pathology classes, so a model trained on them is blind to everything else in the scan. Screener, from IRA Labs with collaborators at the Marchuk Institute of Numerical Mathematics, Lomonosov Moscow State University, and MBZUAI, sidesteps the ceiling by reframing pathology detection as unsupervised visual anomaly segmentation. Pathological tissue is, by definition, rare relative to healthy tissue; a model that has learned what normal anatomy looks like can flag departures from it without ever being told what a lesion is.

The density-based approach to anomaly segmentation is not new — extract dense features, fit a density model to them, score low-density voxels as anomalous. Screener's contribution is to make both halves of that pipeline learned. Feature extraction uses dense self-supervised learning on unlabeled CT rather than features borrowed from a supervised or ImageNet-pretrained encoder, and the conditioning variables that tell the density model where in the body a voxel sits are learned masking-invariant descriptors rather than hand-crafted positional encodings. Trained on more than 30,000 unlabeled CT volumes, the result transfers to held-out cohorts with no fine-tuning at all.

#Key Features

  • No annotations anywhere in the pipeline: Both the descriptor model and the density model train on unlabeled volumes, so coverage is not limited to the pathology classes some dataset happened to mark up.
  • Dense self-supervised descriptors: A DenseVICReg objective yields voxel-level features tuned to CT, which the ablations show beat both ImageNet-pretrained ResNet50 features and a supervised anatomy-segmentation encoder by a wide margin.
  • Learned conditioning: Masking-invariant dense features stand in for sinusoidal or anatomical positional encodings, letting the density model condition on where a voxel is without a hand-designed coordinate system.
  • Normalizing-flow density modelling: Swapping a Gaussian density estimate for a flow lifts voxel AUROC from 0.61 to 0.88 on kidney lesions and 0.71 to 0.93 on liver lesions in the ablation.
  • Distills into a single UNet: The whole pipeline collapses into one UNet, which is what makes the pretrained weights usable as an initialization for supervised fine-tuning.

#Technical Details

Screener has three components, all built on UNet-like architectures: a descriptor model trained with a dense VICReg objective at 32 feature dimensions, a condition model producing masking-invariant dense features, and a density model implemented as a normalizing flow. A distillation step then folds the whole pipeline into a single UNet. Pretraining draws on three public collections — NLST, AMOS, and AbdomenAtlas — totalling more than 30,000 unlabeled CT volumes used without curation or filtering. Evaluation covers four external datasets and 1,820 scans: LIDC, MIDRC-RICORD-1a, KiTS, and LiTS. In the unsupervised setting Screener reaches voxel-level AUROC of 0.96 on LIDC, 0.87 on MIDRC, 0.90 on KiTS and 0.93 on LiTS, against 0.87 / 0.76 / 0.76 / 0.80 for the strongest baseline, a patched diffusion model. Dice scores are reported but systematically understated, because the evaluation masks label only certain pathologies while the model detects all of them. In a supervised fine-tuning setting using just 25 labeled cases per fold, Screener-initialized nnUNet reaches 0.31 Dice on LIDC against 0.21 from random initialization. Code is public under the MIT license.

#Applications

The natural use is triage: sweep a stack of chest or abdominal CT and surface the scans and regions that do not look like normal anatomy, whether or not the finding belongs to a category anyone has curated. That covers incidental findings, rare disease, and the long tail of pathology that supervised detectors never see. As a pretrained backbone it also gives radiology groups a way to build task-specific segmenters from a few dozen annotated cases instead of hundreds.

#Impact

Screener's result — that domain-specific dense self-supervision beats supervised and ImageNet features for medical anomaly detection by a large margin — is a pointed argument about where the value in medical imaging pretraining actually sits. Its limitations follow from the framing: anomaly scores mark what is unusual, not what is clinically significant, and the evaluation cannot cleanly measure precision because the ground-truth masks are incomplete by construction. The preprint has not yet been peer reviewed, and the released repository provides training pipelines rather than downloadable checkpoints.

At a glance

Released
February 2025
Category
Imaging
Organization
Mohamed bin Zayed University of Artificial Intelligence

Links

GitHub RepositoryResearch Paper

Tags

anomaly_detectioncnnfoundation_modelradiologysegmentationself_supervised

Something wrong?

Much of this page is generated or calculated automatically. Flag anything that looks off and we will re-run it.