BlackVoxel models

Models for
medical imaging.

BlackVoxel develops proprietary models for specific imaging tasks. Each page lists the data, training method, test result and current limitations.

Available models

The metrics below come from the test sets named on each page. They are research results, not clinical validation.

DX / CXR 01
Integrated reference

Chest X-ray

Chest finding classification with an attention map and report draft.

0.817 macro-AUROCNIH test
bv-chesxr-v1View model →
DX / MSK 02
Research model

Limb fracture

Fracture classification on limb radiographs with Grad-CAM attention.

0.887 AUROCFracAtlas test
bv-limbfx-v1View model →
MR / BRAIN 03
Benchmark only

Brain MRI tumor

MRI-slice classification across no tumor, glioma, meningioma and pituitary tumor.

0.9991 macro-AUROCcurated benchmark
bv-brainmr-v1View model →
US / BREAST 04
Research model

Breast ultrasound

Classification across normal, benign nodule and malignant nodule on BUSI.

0.972 macro-AUROCthree classes
bv-breastus-v1View model →
CT / HEAD 05
R&D only

Head CT hemorrhage

Multi-label classification of intracranial hemorrhage and five subtypes.

0.932 macro-AUROCsix labels
bv-headct-v1View model →
CT / CHEST 06
Research model

Lung nodule characterization

Binary characterization of a crop centered on a lung nodule.

0.931 AUROCpatient-split test
bv-chestct-v1View model →
US / ABD 07
Research model

Abdominal organ recognition

Classification of ten structures in abdominal ultrasound frames.

0.978 macro-AUROCindependent test
bv-abdus-v1View model →
MG / ROI 08
Research model

Mammography ROI characterization

Benign or malignant classification on CBIS-DDSM finding crops.

0.748 AUROCpatient-grouped test
bv-mammo-v1View model →
MR / SPINE 09
Early benchmark

Lumbar MRI findings

Multi-label classification on one midsagittal lumbar MRI slice.

0.762 macro-AUROCsix labels
bv-lumbarmr-v1View model →

Method

How each model is trained and tested.

  1. 01Task

    We define the input, labels and exclusions.

  2. 02Training

    We train the model or fine-tune a pretrained base.

  3. 03Evaluation

    The checkpoint is selected on validation. Reported metrics come from the test set.

  4. 04Review

    Results are shown in the viewer for physician review.

ABOUT THE METRICS

AUROC reports performance on the test set.

The models are proprietary to BlackVoxel. Public datasets and pretrained bases remain under their original licenses. The metrics are research results and do not represent regulatory clearance.

Contact

Talk to BlackVoxel.

Send a message. We will reply to the email provided.