Desenvolvedores

Two changes on the small-bowel block of the operative report. (1) NEW field: each small-bowel resection now records which PORTION of the small bowel it concerned (duodenum | jejunum | ileum), asked for by the surgeons. (2) DOCUMENTATION CATCH-UP, no data change: small_bowel_details was documented as a single opaque JSON column, but the CSV exporter has always recursed into it — a real export already carried initial_length, resections_X_distance, resections_X_length and remaining_length as separate columns that no codebook row described. Those four columns are now declared, and none of them is renamed, moved or removed. WARNING, positional consumers: portion is inserted after each resection's length, so every column downstream of the first resection SHIFTS — read the CSV by header name, never by position. WARNING, consumers reading by NAME: one column name DISAPPEARS from two kinds of file. In the blank template on /developers and in every RADIOLOGICAL export (CT / PET-CT / MRI / FAPIT), operative_report_visceral_resections_small_bowel_details is REPLACED by the five detailed columns (-1, +5). A SURGICAL export is unaffected on this point and keeps whichever of the two shapes applies (see below).

• operative_report.visceral_resections.small_bowel_details.resections.[X].portion — NEW: portion of the small bowel concerned by resection X, as an untranslated code (duodenum | jejunum | ileum). No portion_fr / portion_en companion column: unlike the anastomosis fields, this value is exported as the raw code only. Empty when the surgeon left the select on "-", and for every report saved before this version. One new column per resection actually recorded (N resections → N new columns).
• operative_report.visceral_resections.small_bowel_details.{initial_length, resections.[X].distance, resections.[X].length, remaining_length} — DOCUMENTATION ONLY, no column change in a real export: these four paths were already emitted by every CSV carrying small-bowel data (CsvExportService flattens the block recursively; only morpheus_details is kept opaque) but the schema declared one aggregate path and the codebook one "JSON" row. They now have their own declaration and description. The aggregate path operative_report.visceral_resections.small_bowel_details is KEPT and documented alongside them, because a surgical export whose block holds no data at all still emits it as its single column (see below). A given file carries one shape or the other, never both.
• COLUMN REMOVED where the block is empty by construction — BLANK TEMPLATE and RADIOLOGICAL exports: the JSON skeleton and the blank CSV on /developers now carry the detailed shape {initial_length, resections:[{distance, length, portion}], remaining_length} in place of the single null leaf, so the column operative_report_visceral_resections_small_bowel_details is GONE from them and the five detailed columns take its place (1 → 5 columns on this block, +4 overall). A radiological export (CT / PET-CT / MRI / FAPIT) has no operative report and ships the untouched skeleton, so it loses that column name and gains the five in exactly the same way. A parser that looks the old name up by header in a radiological file will no longer find it.
• TWO SHAPES on this block in a SURGICAL export, both declared and both documented. (a) Block never filled in — every patient without a small-bowel resection, and every report saved before the block existed: the export emits null for the whole block, a null overwrites the skeleton subtree, and the file carries the single empty column operative_report_visceral_resections_small_bowel_details and NONE of the five detailed ones. (b) Block filled in: the five detailed columns appear (plus 3 more per additional resection) and the aggregate column is absent. Consumers must accept both shapes. Note that (a) is a SURGICAL shape only — a radiological file, having no operative report at all, is in shape (b) with every cell empty.
• THIRD SHAPE RETIRED: up to 3.8 a surgical report whose small-bowel panel had been touched but held no resection (the initial length edited, or a resection added then deleted) exported resections as an empty list, which the CSV flattened into one container column named operative_report_visceral_resections_small_bowel_details_resections. That column no longer exists in any file: the skeleton now supplies resection 0, so this case emits the same five detailed columns as any other filled-in block, all empty. The column was never declared in the schema nor described in the codebook, but it did ship in real files — hence this notice.
• HISTORIC RECORDS re-exported after this release: a report saved before 3.9 stored no portion key at all, and its file carries the portion column EMPTY on every resection — not on the first one only. The exporter rebuilds each resection on the declared key set {distance, length, portion}, so the header of an old record is the same as that of a new one; the block is the last list of objects in this export that was copied verbatim from the stored form data. Do not read an empty portion as "portion not applicable": compare against the meta.version of the file.
• MEDICAL CAVEAT on the pairing of portion and distance: distance is measured from the angle of Treitz, which IS the duodeno-jejunal junction, so it has no meaning for a duodenal resection (the duodenum lies upstream). The form records what the surgeon states and fabricates no consistency between the two fields. Likewise remaining_length is computed as initial_length minus every resected length, and initial_length defaults to 600 cm, the length of the JEJUNO-ILEAL small bowel: a duodenal resection therefore subtracts from a capital that does not contain it. Treat remaining_length with caution on any record carrying a duodenal portion.

Radiological review of the PCI by the validating radiologist. Four reference-data changes: the extraperitoneal inguinal canal moves from region 8 to region 5 (it belongs to the two iliac fossae, 5 and 7); region 2 stops listing the foramen of Winslow twice; the fourth unresectability group is rebuilt from one generic placeholder into the five extraperitoneal regions the surgeon cannot fully visualise; and lateralised iliac nodes are added to the abdominal lymph-node group. No path is added or removed from the schema, but the CSV grows by 26 columns and every column after the first change SHIFTS POSITION — read the CSV by header name, never by position. TWO things keep their name and change meaning: criteria_3_items_0_* (now the retroperitoneum, no longer the generic occult lesion) and pci_details_X_nodules_count (which can now be 4, meaning more than 3).

• pci_details.[2].involved_structures — the duplicate occult "Foramen of Winslow" is removed (the region kept the normal one). The four columns pci_details_2_involved_structures_9_{name_fr,name_en,is_involved,diaphragm_score} disappear; the region drops from 10 to 9 entries. Files exported from 3.3 (when the occult spaces were merged into the anatomic-structure list) to 3.7 carried the foramen twice in region 2, at positions 4 and 9, indistinguishable by label alone: no stable structure identifier is exported.
• pci_details.[8].involved_structures / pci_details.[5].involved_structures — the extraperitoneal "Inguinal canal / Canal of Nuck" moves from region 8 to region 5. The four columns pci_details_8_involved_structures_6_* disappear and pci_details_5_involved_structures_7_* appear. It now sits on the two iliac fossae, region 5 (left) and region 7 (right).
• criteria.[3] (occult_lesions) — the single generic "occult (extraperitoneal) lesion" item is replaced by five named extraperitoneal regions, in this order: 0 retroperitoneum (retroperitoneal fat), 1 abdominal wall, 2 epigastric vessels (lymph node), 3 previous abdominal wall scars and port sites, 4 cardiophrenic and subxiphoid regions. The group grows from 7 to 31 columns. WARNING: criteria_3_items_0_* keeps its name but changes meaning — it now designates the retroperitoneum, not the generic occult lesion. The group key stays occult_lesions.
• criteria.[1] (abdominal_lymph_nodes) — new lateralised "Iliac nodes" item, inserted at its anatomical rank between "Retroperitoneal below the renal pedicle" and "Inguinal nodes". The group grows from 5 to 6 items (+6 columns). WARNING, positional consumers: criteria_1_items_4_* now designates the ILIAC nodes (it used to be the inguinal ones) and the inguinal nodes move to criteria_1_items_5_*. criteria_1_items_4_laterality is the first column of this group to carry right | left | bilateral.
• DOCUMENTATION, no column change: the five leaves name_fr, name_en, precision, laterality and options of criteria groups 1, 2 and 3, plus criteria_1_key and criteria_2_key, are now declared in the schema and described in the codebook. They were already shipped in every CSV (the formatter is generic) but only is_involved was documented, so 72 columns of the current file had no codebook row. The blank template on /developers gains them; no column of a real export is renamed, moved or removed — verified by the CSV header contract test.
• pci_details.[X].nodules.count — unchanged column, EXTENDED VALUES: the form offers "> 3" again next to the exact counts 0-3, encoded as 4. A 4 means "more than 3 implants", not four implants; averaging the column without handling 4 separately now overestimates. No version listed in this changelog (2.0 onwards) ever emitted a value above 3. Files carrying the pre-changelog labels 2.6 / 2.7 may already contain a 4: it meant "more than 3" there too, the original encoding being restored unchanged.

New optional location field on digestive anastomoses: records where a colocolic anastomosis sits along the colon (right / transverse / left), so the operative schema can draw it on the correct segment. Three new columns per anastomosis entry; existing columns are unchanged. Same release: the iso/aniso peristalsis choice is now only offered for a side-to-side (LL) configuration — the peristalsis columns are unchanged but will be null for the other configurations going forward.

• operative_report.reconstructive.anastomoses.[X].location — NEW: coded localization along the colon (right | transverse | left), colocolic anastomoses only; null for other types or when not entered (the schema then infers the segment from the recorded colectomy).
• operative_report.reconstructive.anastomoses.[X].location_fr — NEW: French label of the localization.
• operative_report.reconstructive.anastomoses.[X].location_en — NEW: English label of the localization.
• operative_report.reconstructive.anastomoses.[X].peristalsis — unchanged column, restricted INPUT: the CRO form now offers the iso/aniso choice only for the side-to-side (LL) configuration (it has no meaning for end-to-side/side-to-end/end-to-end). New records under other configurations export null; historic values are kept unless the report is edited again.

Each drain now records where it is exteriorised through the wall: a side (left/right) and a level (upper/middle/lower third). Six new columns per drain entry; existing columns are unchanged. Same release also stops the drain code columns from carrying an untranslated key when the picker was left empty.

• operative_report.drains.[X].exteriorization_side — NEW: coded exteriorisation side (left | right), null when not recorded. There is no midline code: a drain brought out through the laparotomy itself has no value of its own today.
• operative_report.drains.[X].exteriorization_side_fr — NEW: French label of the exteriorisation side.
• operative_report.drains.[X].exteriorization_side_en — NEW: English label of the exteriorisation side.
• operative_report.drains.[X].exteriorization_level — NEW: coded exteriorisation level (upper_third | middle_third | lower_third), null when not recorded.
• operative_report.drains.[X].exteriorization_level_fr — NEW: French label of the exteriorisation level.
• operative_report.drains.[X].exteriorization_level_en — NEW: English label of the exteriorisation level.
• operative_report.drains.[X].type / .location — unchanged columns, corrected VALUES: a picker left on its placeholder was exported as an empty code whose label columns then carried the untranslated key ("OR.drains.types."). Such an entry now exports null in the code column and null in both label columns. In the CSV the code cell was already blank, so what moves there is the label cell, from the literal key to an empty cell.

New optional montage field on digestive anastomoses: records the reconstruction technique when several exist for one anastomosis type (currently the gastrojejunal anastomosis: Roux-en-Y limb vs omega loop). Three new columns per anastomosis entry; existing columns are unchanged.

• operative_report.reconstructive.anastomoses.[X].montage — NEW: coded montage (roux_en_y | omega), null for other anastomosis types or when not entered.
• operative_report.reconstructive.anastomoses.[X].montage_fr — NEW: French label of the montage.
• operative_report.reconstructive.anastomoses.[X].montage_en — NEW: English label of the montage.
• operative_report.reconstructive.anastomoses.[X].type — unchanged column (free text), but the picker now also offers the pancreatico-jejunal anastomosis, so this new label can appear in the data (a Whipple has three anastomoses; the report could previously record only two).

Six CRO sections that were exported as a raw client passthrough (incision, reconstructive, biopsies, biopsies_laparo, surgical_procedures_laparo, planned_treatment_laparo) plus hipec.details are now stable whitelisted structures: they always emit the same documented columns (null when absent) and match ExportSchemaRegistry::SCHEMA one-to-one. This removes the silent drift where real exports carried undocumented columns, omitted documented ones, and varied the column set from one patient to the next.

• operative_report.incision.* — full incision geometry now documented (was only 6 keys): approach/type/laterality with _fr/_en labels, midline_details, port_sites, port_site_other, trocar_config (+labels), trocar_count, mini_laparotomy(+details), robot_assisted(+details), other_description, pre_peritoneal, adhesiolysis, ascites_at_opening, ascites_quantity, comment.
• operative_report.reconstructive.* — anastomoses.[X], stomas.[X], urinary.[X] and matrix.[X] (prosthetic mesh) are now whitelisted object lists with bilingual labels; vascular_comment and global_comment added; UI-only DOM ids are no longer exported.
• operative_report.biopsies.* and operative_report.biopsies_laparo.* — biopsy sites and cytologies are emitted as stable structures; the scalar biopsies_laparo column becomes a structured object.
• operative_report.surgical_procedures_laparo.* and operative_report.planned_treatment_laparo.* — the two laparoscopy scalar columns are replaced by their documented field sets.
• operative_report.hipec.details.* — the HIPEC details sub-object is now a fixed whitelist (pathology, surgery, chemo, temp, duration, flow, volume) instead of a raw passthrough (only temp and duration were declared before).

Unresectability criteria fully rebuilt to the final surgical reference document: 30 criteria across peritoneal disease/organ resections, bulky nodes, distant/extra-peritoneal metastases and occult lesions. Criteria can now carry multi-option enumerations (pancreas head/body/tail, ureter thirds, vessel list, abdominal-wall side, small-bowel segments, duodenum parts), exported per item. The occult-lesions column is now emitted in its own criteria group.

• criteria.[].items.[].options — new per-item field: null when the criterion declares no option set, otherwise the list of selected option keys.
• criteria.[3] (occult_lesions) — the occult-lesions criteria group is now emitted in its own bucket (previously silently folded into non_resecability_reasons); criteria.[].key gains the value occult_lesions.
• Criteria set/grouping changed (data, not schema): the whole unresectability criteria set was rebuilt to the final medical document (22 peritoneal-disease / 5 nodal / 2 metastatic / 1 occult).

Unresectability criteria reworked: right/left laterality is now captured for the criteria that support it (ureter invasion, arterial/venous infiltration, massive diaphragm involvement) and exported as a per-item scalar enum. A "second, third and fourth part of duodenum" criterion is added and "Ureter(s) invasion" moves to the peritoneal-disease column.

• criteria.[].items.[].laterality — new per-item field: null when not applicable, otherwise one of right | left | bilateral.
• Criteria set/grouping changed (data, not schema): new "second, third and fourth part of duodenum" criterion; "Ureter(s) invasion" moved from the extra-peritoneal to the peritoneal-disease group.

PIPAC and HIPEC export now match their documented contract: hipec.performed is emitted (was stored but dropped), and the PIPAC section is a stable, documented structure (was passed through raw) enriched with the computed protocol label.

• operative_report.hipec.performed — now emitted (yes/no); previously documented but never produced.
• operative_report.pipac.* — PIPAC is no longer a raw passthrough. Documented keys: performed, context, chemo_details, protocol, protocol_label_fr, drug, dose, duration, comment. Newly exported: context, chemo_details, drug, duration; protocol_label_fr is now computed (was absent from the raw payload).

Imaging PCI morphology aligned with the "Lexicon of morphological terms on imaging": three MORPHEUS 2 terms (nodule, thickening_plaque, fat_infiltration) whose qualifier/complementary extras now drive auto-scoring. Infiltration is merged into the fat_infiltration term and the dedicated infiltration block is removed (BREAKING).

• pci_details.[X].infiltration.{present,retractile,diffuse_retractile} — REMOVED. Fat infiltration is now reported as the "fat_infiltration" term inside pci_details.[X].nodules.morpheus2 / morpheus_extras / morpheus_details (qualifiers fat_infiltration_q_focal | fat_infiltration_q_diffuse).
• pci_details.[X].nodules.morpheus2 — imaging value set replaced: the 9 legacy imaging terms give way to nodule, thickening_plaque, fat_infiltration. (Surgery terms unchanged.)
• pci_details.[X].nodules.morpheus_extras — new imaging qualifier/complementary values: nodule_q_{micronodule,nodule,mass}, nodule_c_{calcified,mucinous,cystic,scalloping}, thickening_plaque_q_{focal,diffuse}, thickening_plaque_c_{calcified,mucinous}, fat_infiltration_q_{focal,diffuse}, fat_infiltration_c_{calcified,mucinous}.

Residual PCI for incomplete cytoreductions, Aletti Surgical Complexity Score for ovarian cancer, and formal documentation of file metadata.

• scores.pci.residual.{applicable,total,max,regions_vector} — residual PCI after an incomplete cytoreduction (CC-1/2/3).
• scores.aletti_total, scores.aletti_complexity_class, scores.aletti_complexity_label — Aletti SCS (ovarian cancer evaluations only).
• aletti_details[] — per-category Aletti breakdown (ovarian cancer evaluations only).
• meta.generated_at, meta.version, meta.language_codes — file metadata now formally documented in the codebook.

Laparoscopy (PIPAC) workflow fields and patient sex.

• evaluation_meta.patient_sex, evaluation_meta.laparoscopy_chemo_name and related laparoscopy fields.
• operative_report.biopsies_laparo, surgical_procedures_laparo, planned_treatment_laparo.

Reconstructed schema (this version predates the machine-readable schema; rebuilt from the change history).

First versioned codebook of the structured surgery/imaging export.

• Initial published structure: evaluation_meta, scores, pci_details, fagotti, criteria and the full operative_report tree.

Downloads unavailable: this version predates the machine-readable schema and cannot be reliably reconstructed.