Aorta Health Help

This guide explains the control panel options in plain language. The app can run neural-network segmentation, phenotype extraction, or both, then show progress and build clickable reports.

The Basic Idea

Aorta Health takes CT files, optionally creates segmentations with nnU-Net, runs phenotype extraction stages, and saves outputs for each case. The web app is the control panel: it chooses which cases and stages to run, starts or pauses the backend, shows the newest preview, and displays the latest report.

Most users only need the visible File Paths and the main Start/Pause/Resume button. Advanced Configuration is for narrowing cases, choosing stage order, or deciding when old outputs should be reused versus regenerated.

File Paths

DatasetThe dataset name is the folder name for the cases you want to run. If the CT folder points to a parent folder, the app can infer the dataset from the CT subfolder.

CT FolderUse either the dataset-specific CT folder or the parent CT folder. CT files usually end in _0000.nrrd.

Segmentation FolderThis is under Advanced Configuration. Use inference segmentations if Aorta Health should generate masks, or supply your own segmentation folder if masks already exist.

Output LocationThe app writes reports, metadata, logs, and per-case outputs under the standard dataset output folder automatically.

Example: If CT Folder is datasets/Inputs/CTs and it contains one dataset folder called Practice 3 sbuh and 2 avt, the app can fill Dataset with Practice 3 sbuh and 2 avt. If CT Folder already points directly to datasets/Inputs/CTs/Practice 3 sbuh and 2 avt, the dataset name is that folder name.

Start, Pause, Resume

Start Pipeline begins a new run using the current configuration.
Pause Pipeline stops after terminating the currently running stage process. The visible preview/report should stay in place.
Resume Pipeline continues the same paused run using the saved configuration from that run.
Resume is stage-based. If the app died during case 2, stage 3, resume restarts case 2, stage 3 from the beginning.

Stages

1 SegmentationRuns nnU-Net on selected CT files and writes inferred masks into the standard inferred segmentation folder.

2 Phenotype ExtractionRuns the downstream analysis stages using either inferred segmentations or your supplied segmentation folder.

2.1-2.7 SubstagesWhen you choose stages manually, Phenotype Extraction expands into Orientation Review, Skeleton/Bifurcation, Posterior-Anterior Labeling, Branch Analysis, Centerline Analysis, Phenotype Extraction, and Final Visualization.

Advanced Configuration

Segmentation DetailsInference Segmentations means Stage 1 creates masks and phenotype extraction uses those masks. Supply My Own Segmentations means phenotype extraction uses a folder you choose.

Segmentation QualityFast Segmentation is the current default: test-time augmentation off and step size 0.8. High Quality Segmentation enables test-time augmentation and uses step size 0.5. It is slower but more conservative.

ScopeRun all cases, or one specific case ID.

Subset FilterOnly appears for all-case runs. It limits all-case runs by case number range, name/regex pattern, or both.

Case OrderChoose the order cases run in, such as alphanumeric, smallest segmentation first, biggest segmentation first, or oldest completed output first.

Execution OrderPer Case finishes all selected stages for one case before moving on. Per Stage runs one stage across all cases before moving to the next stage.

Stage Rerun RuleControls whether selected stages rerun or are skipped based on existing outputs, stale outputs, or real notebook content changes.

StagesThis is the first advanced subsection. Run all stages, or choose Stage 1 Segmentation and/or Stage 2 Phenotype Extraction. Phenotype Extraction can be expanded into stages 2.1 through 2.7.

Reset Advanced ConfigurationIf advanced settings differ from the default, a red reset bar appears under Advanced Configuration.

Subset Filter Examples

All CasesRuns every matched CT/segmentation pair in the dataset.

Case Number RangeUses the number inside the case ID. Range 2 to 50 includes AVT_02, AVT_25, and SBUH_10.

Name Pattern / RegexMatches case names. Pattern AVT runs only cases with AVT in the name. Pattern ^AVT means the name must start with AVT.

Range + PatternBoth rules must match. Range 2-50 plus pattern ^AVT includes AVT_02 and AVT_25, but excludes SBUH_10.

Stage Rerun Rule Examples

Always RunRuns the selected stages even if outputs already exist. Use this when you intentionally want fresh outputs.

Only If Output MissingRuns a stage only when its expected completion output is missing. Existing outputs are reused.

If Stage Notebook Changed Or Output MissingRuns when the stage notebook content actually changed, or when output is missing. Normal notebook output/autosave noise should not count as a meaningful change.

If Output Older Than ThresholdRuns when the selected stage output is older than the Stale Hours value. For example, 48 hours means outputs older than two days are refreshed.

The rerun rule is applied stage-by-stage. If you select stages 3-5, the rule is checked separately for stage 3, stage 4, and stage 5 for each selected case.

Case Order Examples

AlphanumericRuns cases by name, such as AVT_01, AVT_02, then SBUH_01.

Segmentation SizeRuns by segmentation file size, either smallest to biggest or biggest to smallest.

Sort By Last Completed OutputRuns cases with missing or older selected-stage outputs first. If you select only stage 3, it sorts by stage 3 output. If you select stages 3-5, it sorts by stage 5 output.

Only Cases Older Than ThresholdFocuses on cases whose selected reference output is older than the Stale Hours value.

Reports

The embedded report is the latest dataset-level summary. Case links open case-level summaries. Preview links open the latest visual output for that case/stage.

Resume Incomplete

Use this tab if a run was ended, the browser tab was closed, the app stopped, or the computer shut down. The app looks at saved metadata and restarts from the most recent incomplete stage.