Three-dimensional deep-learning age and sex estimation from the fourth rib: de-identified CT-derived surface dataset and reference implementation

```html<p>De-identified data and the reference implementation accompanying <em>Etli Y., Özdil A., “Three-dimensional deep-learning age estimation from the fourth rib on high-resolution CT: an evaluative likelihood-ratio approach”, Forensic Science International (2026)</em>.</p> <p><strong>Data.</strong> <code>rib_stl_deposit.zip</code> contains 453 de-identified surfaces of the right fourth rib (binary STL, millimetres; 225 men, 228 women; 18.0–99.8 years), segmented from thin-section (1.00 mm) clinical chest CT with AI-assisted three-dimensional segmentation in ITK-SNAP, operator review and correction, and translated to their centroids. The surfaces include any calcified or ossified costal cartilage contiguous with the sternal end. Thirty surfaces re-segmented by the same operator about one month later (<code>RIB_XXXX_repeat.stl</code>) support the intra-observer analysis. <code>rib_case_list.csv</code> gives <code>case_id</code>, <code>sex</code> and <code>age_years</code> — the only metadata — and <code>split.json</code> the frozen development/test partition (362/91) used in the article. File numbering was randomly permuted; no names, dates or other identifiers are present (see <code>deidentification_summary_public.md</code>).</p> <p><strong>Code.</strong> <code>rib_age_estimation_zenodo_v1.0.1.zip</code> is the complete single-module pipeline: canonical pose in the scanner's patient frame; four representations (metric multi-view RGB-D renderings and a point cloud of the sternal end, a point cloud of the complete rib, twenty morphometric scalars); sex-specific networks trained by label-distribution learning over one-year age bins; temperature-scaled calibration, de-priorised likelihoods and likelihood ratios for age thresholds with Cllr, Cllr_min and Tippett diagnostics; sex estimation from the same inputs; intra-observer repeatability. MIT licence (inside the archive).</p> <p><strong>Version 1.0.1</strong> replaces the code archive only: the export of the per-case repeatability table (<code>repeatability.csv</code>) mis-aligned its <code>case_id</code> and <code>dice_rib</code> columns after a skipped pair in v1.0.0; ages, sexes, the other overlap metrics, the predictions and every summary statistic were unaffected. Skipped pairs are now written to <code>repeatability_skipped.csv</code>. The data files are byte-identical to version 1.0.0. See <code>CHANGELOG.md</code> inside the code archive and the version history in <code>README.md</code>.</p> <p><strong>Ethics.</strong> Approved by the Non-Interventional Clinical Research Ethics Committee of Van Yüzüncü Yıl University (decision 2025/09-11, 19 September 2025); conducted in accordance with the Declaration of Helsinki; individual informed consent waived for this retrospective use of anonymised, routinely acquired images.</p> <p><strong>Licences.</strong> Data: CC BY 4.0 (<code>LICENSE.txt</code>). Code: MIT (<code>LICENSE</code> inside the code archive). Please cite this record and the article.</p>``` ---