Presets¶

Reusable processing configuration, stored as one JSON file per preset under ~/.optari/config/presets/<category>/.

Presets aren't part of the Settings dialog. Instead, each category is managed from its own dock's Save Preset / Remove Preset controls.

ROI presets (`presets/roi/`)¶

A saved ROI template. The geometry block is the same representation OPTARI uses for the saved analysis table and for HDF5 export: vertices in PATATO coordinates, metres, origin at the image centre, so the shape does not depend on the field of view it was drawn on. tissue_class is the preset's location (shown as Location in the Annotation dock): the segmentation class an auto preset is anchored to.

source_fov_m records the field of view the template was drawn on, in metres. It is needed to re-place the template proportionally when it is applied to a scan with a different field of view,

placement describes the way the template is positioned:

Value	Meaning
`static`	Put it at its saved coordinates, scaled to the scan's field of view.
`auto`	Anchor it onto the segmentation class named by `geometry.tissue_class`. Needs a segmentation map.

top_margin_mm is how far below the class's top edge (at the image's horizontal centre) an auto preset's top is placed, horizontally centred. The Save Preset popup prefills it with where the ROI sits when saved, and it can be edited there. 0 places the ROI's top right on the class's top edge.

Placing an auto preset on a scan without a matching segmentation fails with an error, both in the Annotation dock and in a batch run.

clinical_ellipse_10x2mm.json

{
  "description": "Ellipse of size 10x2mm similar to 2nd UK PAD study",
  "created": "2026-04-08T12:49:27.693948+00:00",
  "source_fov_m": [0.04, 0.04],
  "placement": "static",
  "geometry": {
    "verts_m": [[-0.005, 0.0], [0.005, 0.0], [0.005, -0.002], [-0.005, -0.002]],
    "kind": "ellipse",
    "tissue_class": "muscle"
  }
}

See ROI Annotation for how to create these from the GUI.

Reconstruction presets (`presets/reconstruction/`)¶

Reconstruction algorithm and filtering parameters, passed through to PATATO and DeepMB.

backproject_clinical.json

{
  "FILTER_HIGH_PASS": null,
  "FILTER_LOW_PASS": 7500000.0,
  "IRF": true,
  "HILBERT_TRANSFORM": true,
  "ENVELOPE_DETECTION": false,
  "INTERPOLATE_TIME": 3,
  "INTERPOLATE_DETECTORS": 2,
  "PREPROCESSING_ALGORITHM": "Standard Preprocessor",
  "RECONSTRUCTION_FIELD_OF_VIEW_X": 0.04,
  "RECONSTRUCTION_FIELD_OF_VIEW_Z": 0.04,
  "RECONSTRUCTION_NX": 400,
  "RECONSTRUCTION_NZ": 400,
  "RECONSTRUCTION_PARAMS": {},
  "RECONSTRUCTION_ALGORITHM": "Reference Backprojection",
  "RECONSTRUCTION_SPEED_OF_SOUND": 1500
}

Unmixing presets (`presets/unmixing/`)¶

Wavelength range, chromophore selection, and derived-parameter options.

haemoglobin.json

{
  "RESOLUTION_REDUCE": 1,
  "WAVELENGTH_RANGE": [700, 900],
  "SPECTRA": ["Hb", "HbO2"],
  "SO2": true,
  "THB": true,
  "SUFFIX": ""
}

SPECTRA entries reference PATATO's reference chromophore spectra (H2O, HbO2, Hb, ICG, Lipid, Melanin) for a wavelength range from 700 to 1000nm. (Upcoming: Spectra up to 1300nm)

Segmentation presets (`presets/segmentation/`)¶

Which model/class to run, and how to place an ROI from the resulting mask (see Segmentation Models for the model registry referenced by model_id).

muscle.json

{
  "model_id": "c_unet_gastrocnemius-transverse",
  "selected_class_ids": [3],
  "roi_class_id": null,
  "roi_shape": "ellipse",
  "roi_width_mm": null,
  "roi_height_mm": null,
  "roi_top_margin_mm": null
}

Batch presets (`presets/batch/`)¶

An entire analysis applied to a whole dataset. Unlike the other categories a batch preset holds no processing settings of its own, and isntead it references the presets above. See the Batch Processing.

Batch presets are edited in their own window (OPTARI → Batch Processing…), which shows the JSON, Save Preset / Remove Preset, and a Validate button that checks the plan against the dataset.

clinical_muscle_roi.json

{
  "description": "Reconstruct, segment, unmix and place the clinical muscle ROI.",
  "steps": {
    "reconstruction": "backproject_ithera",
    "segmentation": "muscle_poly",
    "unmixing": "haemoglobin",
    "roi": "clinical_ellipse_10x2mm_Muskel1"
  },
  "source": null,
  "frame": "spectral_motion",
  "all_frames": false,
  "measure": {
    "layers": "analysis",
    "all_channels": true
  },
  "outputs": {
    "xlsx": true,
    "hdf5": false,
    "overlay_png": true,
    "overlay_layer": null
  }
}

`steps`¶

Each entry is a preset name only. Omit a step to skip it.

Key	Preset category	Notes
`reconstruction`	`presets/reconstruction/`	Runs first. Omit to work on a reconstruction already in the file.
`segmentation`	`presets/segmentation/`	Required by an ROI preset whose `placement` is `auto`.
`unmixing`	`presets/unmixing/`	Wavelengths resolve per scan, so a scan missing them fails on its own.
`roi`	`presets/roi/`	The ROI placed and measured. Its `placement` decides static or auto. The value `"from_scan"` places nothing and measures the ROIs the scan already holds (e.g. iLabs `.iannotation`/`.iROI`, or saved in an earlier OPTARI export), each on the frames it was drawn on. It implies `all_frames`.

An empty "steps": {} with outputs.hdf5 on is a pure vendor-to-OPTARI-HDF5 converter.

`source`¶

A layer-name prefix picking the one reconstruction the run analyses. One batch run works on one reconstruction, which produces one unmixing, so a scan holding several reconstructions still produces one analysis.

Value	Meaning
`null` with a `reconstruction` step	The reconstruction this plan just produced.
`null` with no `reconstruction` step	The scan's default OA layer (see `DEFAULT_PA_LAYER`). Flagged as a warning, since it is implicit.
`"Recon: iThera"`	The layer whose name starts with this. A scan without one fails and the run continues.

`frame`¶

The anchor frame: where a static ROI's shape is drawn, and which frame the overlay PNG and the report's analysis_frame column come from. It does not decide how many frames get analysed, which is selected by all_frames below.

Value	Meaning
`"spectral_motion"` (default)	That scan's frame with the least motion during its set of wavelength (relevant for unmixing). A single-wavelength scan falls back to `"frame_motion"`. Requires ultrasound. A raw time series (eg. IPASC) has none and falls back to frame 0.
`"frame_motion"`	That scan's frame that differs least from its previous and next frame.
`"acquisition_start"`	The frame where the operator pressed record in ithera scans, (shown as sweep 1 in iLabs). Scans that do not record it fall back to frame 0.
`"roi"`	The first frame holding one of the scan's ROIs, so the overlay shows them. Meant for `"roi": "from_scan"`, where every scan's ROIs can sit on a different frame. Scans without ROIs fall back to frame 0.
an integer	That frame number. A scan without it fails.

`all_frames`¶

false (default) runs every step and the measurement on the frame above only. true runs reconstruction, segmentation, unmixing and measurement on every frame. The ROI is placed on each frame (re-anchored per frame for auto placement) and measured there. frame still picks the overlay/report anchor. "roi": "from_scan" always implies true.

`measure`¶

How wide the measurement reaches. Mirrors the Include all … boxes in the Annotation dock.

Key	Values	Meaning
`layers`	`"analysis"` (default), `"all_pa"`	`analysis` measures the `source` reconstruction plus what this run unmixed from it. `all_pa` measures every OA layer in the scan, including reconstructions the plan did not make.
`all_channels`	`true` (default), `false`	Every wavelength of the reconstruction, or just the default one. Derived layers (unmixed, sO₂, THb) are always measured on every channel.

`outputs`¶

Key	Values	Meaning
`xlsx`	`true` (default), `false`	Write `batch_roi_table.xlsx`, the measurement table.
`hdf5`	`false` (default), `true`	Write each scan to `hdf5/<Study>/<Scan>.hdf5`.
`ipasc`	`false` (default), `true`	Write each scan's raw time series to `ipasc/<Study>/<Scan>_ipasc.hdf5` (see Exporting Data).
`overlay_png`	`true` (default), `false`	Write one viewer screenshot per scan to `overlays/`.
`overlay_layer`	`null` (default), a layer-name prefix	Which layer the overlay shows. `null` uses the last one the run produced (sO₂ if unmixing made it, else Unmixed, else the reconstruction). A prefix matching nothing fails that scan.

batch_report.xlsx and the run log are always written.

Any preset is just a JSON file that can be copied into another machines ~/.optari/config/presets/<category>/ to reuse the same configuration across workstations.

Presets¶

ROI presets (presets/roi/)¶

Reconstruction presets (presets/reconstruction/)¶

Unmixing presets (presets/unmixing/)¶

Segmentation presets (presets/segmentation/)¶

Batch presets (presets/batch/)¶

steps¶

source¶

frame¶

all_frames¶

measure¶

outputs¶

Sharing presets¶