# Patient-specific implant design lineage: synthetic field guide

Version 1.1. Published October 23, 2026 by Operavue. Adds a local manifest checker to the original synthetic example.

This educational example contains six invented export records across four invented cases. It includes no scans, segmentations, geometry, patient identifiers, approval documents, or clinical results. Every artifact and review reference is fictional and does not resolve to an included file. The labels are original research-planning fields, not a DICOM schema or clinical standard.

Worked example: https://operavue.com/insights/ct-to-patient-specific-implant-datasets

## Fixed research task

For this exercise, the target is an implant surface mesh documented as approved for its case at one fixed export snapshot. It is not a study of clinical safety, manufacturing suitability, or the uniquely correct implant shape. A record describing approval for one design version does not approve another version. All approval decisions in this exercise are fictional.

## Fields

| Field | Meaning |
|---|---|
| `case_ref` | Opaque fictional case reference; not proof of identity linkage or de-identification. |
| `source_series_ref` | Imaging series the record says informed the design. A real release must resolve this to the exact delivered, authorized artifacts. |
| `anatomy_ref` | Referenced anatomy representation and its version; separate from the implant. |
| `requirements_ref` | Version of the design requirements available for this revision. No clinical parameters are supplied in this example. |
| `design_ref` | Unique reference to the proposed implant design revision. |
| `design_version` | The revision number within this invented case sequence; not an approval ranking. |
| `predecessor_ref` | Previous design reference; blank when this example declares no predecessor. A real unknown should be recorded explicitly. |
| `export_kind` | `implant_mesh` or `viewer_screenshot` in this exercise. It describes what the export is said to contain. |
| `export_format` | Declared representation: STL or PNG. A format name does not establish file validity or export rights. |
| `geometry_units` | Declared distance units for the geometry. Blank in B because the units are missing; blank in D because no geometry is supplied. |
| `geometry_frame_ref` | Coordinate frame of the declared geometry; blank in D. |
| `image_frame_ref` | Coordinate frame used for the source image mapping; blank in D because this screenshot record does not supply a spatial mapping. |
| `mapping_status` | `shared_frame_declared`, `mapping_not_supplied`, or `no_geometry_supplied`. These declarations are not independent alignment verification. |
| `review_record_ref` | Fictional supporting review record. Real records need the reviewer role, purpose, decision, and time or snapshot context. |
| `reviewed_design_ref` | The exact design version named by that review record. It may differ from this row's export. |
| `review_decision` | Decision about `reviewed_design_ref`, not automatically about this row's `design_ref`. `approved_for_case` is a fictional case-level decision and is not regulatory approval. |
| `split_group_ref` | Example grouping key. The three A revisions stay together. Real grouping must also address other known shared patients, anatomy, and source material. |
| `example_interpretation` | Educational reading of the stated circumstances, not an automated verdict or clinical recommendation. |

## Reading the six records

- **A-v1:** the provided review record concerns A-v2. Keep v1 as history; it is neither the approved reference under this task nor a proven clinically incorrect design.
- **A-v2:** the delivered design reference matches the review record. Units and a shared frame are declared. This is consistency of declarations only; the example supplies no real artifacts or evidence that could pass a release review.
- **A-v3:** a newer draft exists, but the review record still names A-v2. Selecting the highest version number would silently substitute an unreviewed export for this task's reference.
- **B-v1:** the geometry is said to be an implant mesh, but distance units are missing. Do not infer millimeters from the filename, application, or other cases.
- **C-v1:** geometry and image frames differ, and no mapping is supplied. Request the mapping with its direction, units, and conventions, then check alignment in the appropriate review environment.
- **D-v1:** a PNG viewer screenshot is listed. The implant mesh requested by this task is not supplied, regardless of whether a design review record exists.

## Run the local manifest checker

Download `check_implant_lineage.py` and `implant-design-lineage-example.csv` from the worked-example page. Save them in the same folder, then run:

```sh
python3 check_implant_lineage.py implant-design-lineage-example.csv
```

Python 3 and its standard library are sufficient. The script reads the named CSV, prints JSON to standard output, and exits. It makes no network requests, writes no files, and does not open any referenced scans, designs, or review documents. The report contains fixed finding codes, column names, and CSV record numbers, with the header counted as record 1. It does not echo paths, case references, or free text. Record numbers differ from physical line numbers when a quoted cell contains a line break.

The published example must return exit code **1**, with six records, five records with findings, and these five findings:

| CSV record | Example design | Finding code |
|---|---|---|
| 2 | A-v1 | `reviewed_design_mismatch` |
| 4 | A-v3 | `reviewed_design_mismatch` |
| 5 | B-v1 | `missing_geometry_units` |
| 6 | C-v1 | `unresolved_mapping` |
| 7 | D-v1 | `missing_implant_mesh` |

Record 3, A-v2, has no declaration finding. That result does not approve a design or release: there are no actual artifacts to inspect in this synthetic example. A-v1 and A-v3 can remain useful history even though they are not the review-named target for this fixed exercise.

Exit **0** means no declaration findings under these rules; exit **1** means findings; exit **2** means invalid input prevented a reliable check. Invalid-input reports contain a fixed error code, not a partial acceptance result. `python3 check_implant_lineage.py --help` prints the supported schema and limits as JSON and exits 0.

## Checker input contract

The checker requires all 18 columns listed above, exactly once, in any order. It accepts UTF-8, an optional byte-order mark, and quoted CSV cells; it trims whitespace at cell edges. Extra or missing columns, duplicate headers, blank or ragged records, no data records, and duplicate nonblank `design_ref` values cause exit 2. Input limits are 2 MiB, 10,000 data records, and 4,096 characters per field.

- Required references are `case_ref`, `source_series_ref`, `anatomy_ref`, `requirements_ref`, `design_ref`, `review_record_ref`, `reviewed_design_ref`, and `split_group_ref`. Each must be an opaque token of 1 to 128 ASCII letters, digits, underscores, periods, colons, or hyphens, starting with a letter or digit. Blank values and reserved unknown markers such as `unknown`, `N/A`, `null`, and `TBD` produce findings.
- `design_version` is an integer from 1 through 999999999, without leading zeros. A blank `predecessor_ref` declares no predecessor; it does not mean unknown. A supplied predecessor must refer to a design present in the same case, with an earlier version. Cycles and duplicate versions within a case produce findings. This exercise does not model branching design histories.
- `implant_mesh` requires `STL`; `viewer_screenshot` requires `PNG`. These are the exercise's supported representations, not a list of every valid research format. A screenshot always receives `missing_implant_mesh` under this target rule.
- Mesh distance units must be `mm`, `cm`, `m`, or `um`. The checker neither converts nor verifies dimensions. Both frame references must be present and valid. Only `shared_frame_declared` with identical frame references can have no mapping finding. `mapping_not_supplied` remains unresolved. Screenshot rows require `no_geometry_supplied` and blank geometry units and frame references. Supplied transforms are outside this schema.
- `review_decision` must be `approved_for_case` for this fixed exercise. The reviewed design must be present in the same case and match the row's design for that row to have no review finding. A reused review-record reference must name the same reviewed design and decision in every occurrence.
- Revisions sharing a case reference must share a split-group reference. This does not establish independence between different cases or discover shared patients or anatomy.
- `example_interpretation` is ignored by the checker. Changing this text cannot resolve a finding.

Blank, unknown, and unsupported required declarations produce findings; they are not treated as a clean result. Resolve discrepancies using source evidence. Do not relabel an export, invent units or a review record, or declare frames identical merely to suppress a finding. If your task uses another representation, review state, transform, or lineage structure, adapt the schema and checks deliberately rather than forcing it into this example.

## Adapting the example

Agree on the target, input cutoff, representation, and treatment of incomplete records first. Add resolvable artifact references, integrity checks, documented derivation steps, any required transform records, reviewer evidence, and authorized-use scope in the governed project environment. Preserve the distinction between missing information and information that is not applicable.

Neither the CSV nor the checker can check geometry, anatomical correspondence, clinical fit, strength, manufacturing readiness, study independence, privacy, or licensing. Matching declarations do not verify that referenced artifacts or approvals exist. Access to imaging, a viewer, or an approved design does not establish rights to export or reuse every artifact. Real patient-level manifests and proprietary designs do not belong in public repositories, business inquiry forms, or ordinary email.

Prepared with AI assistance and reviewed for consistency with the article. No clinical review or endorsement is claimed. The article links the primary sources and explains their scope.