Apollo

CT analysis for research, on your own computer.

Apollo puts published CT models behind one workflow—import, quality control, queue, results, export—and runs it on the machine in front of you.

Apollo does not upload your scans.

Images, metadata, and results are processed on your own computer, and exports are saved where you choose. There is no account, and license keys are checked without a network connection.

Apollo does need an internet connection to install a model and to download its weights the first time it runs. TotalSegmentator, which several analyses use, also reports each run to its authors by default, with details such as the task, platform, software versions, and a random installation ID, but nothing from the scan.

From a folder of DICOM to an exported table.

  1. Import

    Add folders, ZIP archives, or individual files, and Apollo sorts the DICOM inside into studies and series.

  2. Check

    Every series is quality-checked on import, with a plain explanation of each finding.

  3. Choose the analyses

    Apollo shows which analyses can run on each series, and why the others cannot.

  4. Run

    Queue one scan or a whole cohort; a scan that fails is flagged and the rest carry on.

  5. Export

    Export the results to CSV, Excel, or JSON.

Published models, and measurements Apollo computes itself.

Apollo runs 18 analyses. Those from open-source projects are listed with their license and source. Coverage is deepest in the chest, and reaches head, abdomen, spine, and pelvis studies as well.

  1. Sybil

    Risk predictionChestMITSource

  2. Lung segmentation

    SegmentationChestApache-2.0Source

  3. Lung densitometry

    QuantificationChestComputed by Apollo from Lung segmentation

  4. Radiomic features

    Feature extractionChestBSD-3-ClauseSource

  5. Anatomy segmentation

    SegmentationAny regionApache-2.0Source

  6. Lung nodule segmentation

    SegmentationChestApache-2.0Source

  7. Pulmonary vessels and airways

    SegmentationChestApache-2.0Source

  8. Pleural and pericardial effusion

    SegmentationChestApache-2.0Source

  9. Intracerebral haemorrhage

    SegmentationHeadApache-2.0Source

  10. Liver segments (Couinaud)

    SegmentationAbdomenApache-2.0Source

  11. Vertebral bodies and discs

    SegmentationSpine, Chest, Abdomen, PelvisApache-2.0Source

  12. Vertebral bone density

    QuantificationSpine, Chest, Abdomen, PelvisComputed by Apollo from Vertebral bodies and discs

  13. Liver attenuation

    QuantificationAbdomen, ChestComputed by Apollo from Anatomy segmentation

  14. Aortic calibre

    QuantificationChest, AbdomenComputed by Apollo from Anatomy segmentation

  15. 3D surfaces

    SegmentationAny regionComputed by Apollo from Anatomy segmentation

  16. Nodule location

    QuantificationChestComputed by Apollo from Anatomy segmentation and Lung nodule segmentation

  17. Acquisition geometry

    QuantificationAny regionComputed by Apollo

  18. Renal cysts

    SegmentationAbdomenApache-2.0Source

Each result notes how it was produced.

Every result records the model and its version, the quality-control status of its series, and the DICOM slices it was computed from.

What Apollo does not claim.

Apollo is research software. It is not FDA-cleared and does not produce management recommendations. Sybil’s attention overlays show where the model looked, not that a region is cancer. Density and attenuation values depend on the scanner and protocol, so compare them within a consistently acquired cohort. Checking models against their authors' reference outputs is work in progress, starting with Sybil.

Free to install, on Apple Silicon Macs and Windows.