fix(vs-pacs)!: send only pr2mask columns to redcap

research/vestibular_schwannoma/deployment/pacs/README.md

@@ -106,45 +106,18 @@ least 32768 and values outside it lowered to at most 32767: the blue region is e
 mask, and yellow marks voxels with some foreground probability outside it. Existing `fused`, `fused_vote_map`, `reports`, `labels`, and `redcap` directories are rejected
 to prevent overwrites.
 
-### Recoverable masks in REDCap JSON
-
-The final output now includes `redcap/<mask-series-UID>/output.json` and
-`output_data_dictionary.zip`, using pr2mask's EAV row format and instrument.
-`redcap` is an owned output directory and must not already exist.
-The exporter reads the **written DICOM mask**, matches it to the source images by
-geometry, and stores the full binary mask once in the model's permanent `pr2mask` repeat instance.
-It also handles an empty measurement list using the source PatientID and
-ReferringPhysicianName (`EventName:` prefix removed), matching pr2mask routing.
-
-Added fields: `vs_mask_json` and `vs_measurements_json` (Notes), `vs_model_type`, `vs_bundle_sha256`,
-`vs_prediction_id`, `vs_deployment_version`, `vs_tta`, and `vs_created_at`.
-The existing data dictionary is extended; measurement definitions and OriginID
-are retained. Compression happens in memory, so the mask is JSON text rather
-than a separate binary attachment. There is no direct REDCap uploader.
-
-`vs_mask_json` is a JSON string with schema version 1:
-
-- Binary voxels: C-order `(slice, row, column)`, foreground 1, little-order bits,
-  gzip-compressed and Base64-encoded (`packbits-gzip-base64-v1`).
-- Shape, foreground count, and SHA-256 of the decoded C-order uint8 voxel bytes.
-- DICOM LPS geometry: `pixel_spacing` in row/column order,
-  `image_orientation_patient`, and `image_positions_patient` in encoded slice
-  order. **All lengths are millimetres by schema convention; no units field.**
-  Slice order follows increasing position along the DICOM orientation normal.
-  Per-slice positions preserve irregular spacing; SliceThickness is not used
-  as a substitute for the distance between slice centres.
-- Source Study/Series/Frame of Reference and ordered SOP UIDs, mask series UID,
-  model type, member IDs, bundle hash, deployment version and TTA.
-- Stable prediction identity (hash of mask, geometry and provenance) and UTC
-  export timestamp. The timestamp is excluded from the prediction identity.
-
-Recover voxels with `redcap_output.decode_mask(json.loads(row['value']))` for the
-`vs_mask_json` row. For voxel `[s, r, c]`, its LPS position is
-`position[s] + r * pixel_spacing[0] * orientation[3:6] +
-c * pixel_spacing[1] * orientation[0:3]`.
-Use the same spatial grid as the reference before computing Dice. Distance
-metrics must respect geometry; irregular grids require physical-coordinate
-surface distances or a documented resampling step.
+### REDCap output
+
+The final output includes pr2mask's `redcap/<mask-series-UID>/output.json` and
+`output_data_dictionary.zip`, in pr2mask's EAV row format and instrument. `redcap` is an
+owned output directory and must not already exist. There is no direct REDCap uploader.
+The exporter sends only pr2mask's own columns, in the model's permanent `pr2mask` repeat
+instance. An empty measurement list is routed with the source PatientID and
+ReferringPhysicianName (`EventName:` prefix removed), matching pr2mask routing. The mask
+itself goes to the PACS as the `labels` series.
+
+Up to 20261026T134845Z the exporter also sent eight `vs_*` fields, among them the mask as
+`vs_mask_json`. They are no longer sent.
 
 ### Stable model destinations and overwrites
 
@@ -164,30 +137,26 @@ All fields for a model share its repeat instance. Within the same record/event:
 - The same bundle overwrites its previous result, including when TTA, deployment
   version, timestamp or the predicted mask changes.
 - A different bundle uses a different instance, including different weights of
-  the same architecture. `vs_prediction_id` remains provenance, not a routing key.
-
-pr2mask's per-region measurements are preserved in `vs_measurements_json`, a JSON
-list of `{redcap_repeat_instance, field_name, value}` entries. The instance inside
-this list is the original **region** number, not the model destination. Regions are
-not emitted as separate top-level repeat instances, which would collide with the
-model instances. Instead, pr2mask's own measurement columns in the model instance
-hold the **largest** region (most voxels; ties keep the lower region number), and the
-list holds every region. pr2mask numbers regions in scan order, so the largest is not
-always region 1. An empty mask has no regions, so pr2mask lists no columns; the exporter
-then takes them from `pr2mask_dictionary.csv` and sends them blank, so the importer must
-overwrite blanks to clear values from an earlier run.
-
-Since the Fiona base of 2026-10-19, pr2mask also stores each region's mask in seven
-`mask_*` fields. These fill the columns for the largest region but are left out of
-`vs_measurements_json`: the mask alone is about 84,000 characters for a 512×512×120 scan,
-and REDCap cuts values longer than 65,535 characters, so it only decodes for small volumes.
-`vs_mask_json` holds the lossless mask in about 22,000 characters.
-
-`pr2mask_dictionary.csv` holds the `instrument.csv` rows (without the `vs_*` fields) of a
-non-empty run of the packaged pr2mask. When the Fiona base changes, regenerate it from a
-smoke test and check that a non-empty run lists the same columns.
-
-Install the extended dictionary in the receiving REDCap project and qualify a
+  the same architecture.
+
+pr2mask writes each region as its own repeat instance (1, 2, 3, ...), which would collide
+with the model instances, so regions are not sent as separate instances. pr2mask's columns
+in the model instance hold the **largest** region (most voxels; ties keep the lower region
+number). pr2mask numbers regions in scan order, so the largest is not always region 1. An
+empty mask has no regions, so pr2mask lists no columns; the exporter then takes them from
+`pr2mask_dictionary.csv` and sends them blank, so the importer must overwrite blanks to
+clear values from an earlier run.
+
+Since the Fiona base of 2026-10-19, pr2mask also stores the mask in seven `mask_*` fields,
+which fill the columns for the largest region. `mask_base64_gzip_compressed` is about 84,000
+characters for a 512×512×120 scan, and REDCap cuts values longer than 65,535 characters, so
+it only decodes for small volumes. The full mask is in the PACS `labels` series.
+
+`pr2mask_dictionary.csv` holds the `instrument.csv` rows of a non-empty run of the packaged
+pr2mask. When the Fiona base changes, regenerate it from a smoke test and check that a
+non-empty run lists the same columns.
+
+Install pr2mask's dictionary in the receiving REDCap project and qualify a
 PACS-to-REDCap round trip before use. The importer must honor the supplied numeric
 repeat instance and update existing values. Live importer behavior and field limits
 have not been verified. The record/event must already identify the intended case/scan;

research/vestibular_schwannoma/deployment/pacs/pacs_inference.py

@@ -29,7 +29,7 @@
     TTA_FLIP_AXES,
     ParallelEnsemblePatchInferenceEngine,
 )
-from redcap_output import write_redcap_mask, model_repeat_instance
+from redcap_output import write_redcap_output, model_repeat_instance
 
 
 SW_BATCH_SIZE = 1
@@ -286,7 +286,7 @@ def _run_postprocessing(
                 check=True,
             )
     _tag_pr2mask_labels(work_dir, deployment, use_tta=use_tta)
-    write_redcap_mask(work_dir, input_dir, deployment, version=version, use_tta=use_tta)
+    write_redcap_output(work_dir, input_dir, deployment)
     _copy_final_outputs(work_dir, output_dir)
     _publish_log(log_path, output_dir)