--- geometry: - scale=0.75 --- ## General information Data Set Title: "Load-to-failure prediction through CT-based subject-specific FE models". Data Set Author/s: - Julia Aleksandra Szyszko (Alma Mater Studiorum - University of Bologna), ORCID 0000-0001-8750-4719; - Alessandra Aldieri (Alma Mater Studiorum - University of Bologna), 0000-0002-2397-3353; - Fabio Baruffaldi (Medical Technology Lab, IRCCS Istituto Ortopedico Rizzoli), ORCID: 0000-0001-8784-5652; - Marco Viceconti (Alma Mater Studiorum - University of Bologna), ORCID 0000-0002-2293-1530. Data Set Contributor/s: - Daniele Conti (Medical Technology Lab, IRCCS Istituto Ortopedico Rizzoli), ORCID 0000-0001-7790-1921; - Marilina Amabile (Medical Technology Lab, IRCCS Istituto Ortopedico Rizzoli), ORCID 0000-0002-2534-9036; Data Set Contact Person/s: Alessandra Aldieri (Alma Mater Studiorum - University of Bologna), ORCID 0000-0002-2397-3353, alessandra.aldieri@unibo.it. Data Set License: The provided data are provided under the Creative Commons Attribution 4.0 International License (CC BY 4.0). To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. Publication Year: 2026 Project Info: This research was co-funded by the Italian Complementary National Plan PNC-I.1 "Research initiatives for innovative technologies and pathways in the health and welfare sector" D.D. 931 of 06/06/2022, "DARE - DigitAl lifelong pRevEntion" initiative, code PNC0000002, CUP: B53C22006230001. ## Dataset description The dataset comprises 17 CT scans of two cadaveric femurs submerged in water and 17 CT scans of the European Spine Phantom (ESP), acquired at the Istituto Ortopedico Rizzoli (IOR), Bologna, Italy. The femur and phantom CT scans are organised in the following folders: - `CTdata\Femurs\F_PXXX` - `CTdata\ESP_Phantom\ESP_PXXX` where PXXX identifies the CT acquisition protocol: - P\[mA\]: 1.25 mm reconstructed slice thickness with fixed tube current (e.g., P80). - PMOD and PMOD15: 1.25 mm reconstructed slice thickness with automatic tube-current modulation. PMOD uses the scanner default Noise Index (NI = 36.18), whereas PMOD15 uses a target NI of 15. - P\[mA\]fine, PMODfine, PMOD15fine, and PMOD20fine: 0.625 mm reconstructed slice thickness. PMOD15fine and PMOD20fine use target NI values of 15 and 20, respectively, whereas PMODfine uses the scanner default NI of 51.17. The CT acquisition parameters for all protocols are reported in `CTdata\CTdata_Phantom.csv`. ## Images calibration The femur CT scans were calibrated offline using CT scans of the European Spine Phantom (ESP) acquired at IOR. For each femur scan, the phantom scan acquired using the corresponding CT acquisition protocol was used for calibration. A descriptive document for the European Spine Phantom is available from the manufacturer: [link](https://www.qrm.de/en/european-spine-phantom-esp). Seventeen phantom scans were used for calibration. For each phantom scan, the mean HU values of the three hydroxyapatite inserts were calculated and used to derive a linear calibration relationship between CT attenuation and density. The calibration coefficients for all 17 CT protocols are provided in `CTcalibrationPhantom\CalibrationLines.csv`. The calibration relationship is defined as _Density_ (mg/cm3) = _a + b * HU_ where (_a_) and (_b_) are the intercept and slope coefficients, respectively. The mean and standard deviation of the HU values measured in the three ESP inserts for each scan are provided in `CTcalibrationPhantom\Phantoms.xlsx`. ## FE model development and simulation workflow All CT scans were segmented using a semi-automatic segmentation method [[Aldieri2024]] to extract the left and right femur geometries. A 10-node tetrahedral finite element mesh with a nominal 2 mm edge size was generated from the segmentation of the P80 CT scan using the Octree algorithm implemented in Ansys ICEM CFD (release 2019R3, Ansys Inc.). The same mesh was subsequently used for the FE models generated from all CT acquisition protocols. CT-based heterogeneous linear elastic material properties were assigned element-wise using Bonemat (build 152, Bioengineering and Computing Laboratory, Istituto Ortopedico Rizzoli) [[Taddei2007]]. Each FE model was simulated under 28 different combinations of load and boundary-condition orientations representing different impact poses. For each simulation, the load-to-failure was calculated based on the maximum and minimum principal strains and the corresponding failure strain thresholds [[Altai2019]][[Bhattacharya2019]][[Aldieri2023]]. The dataset provides the FE results for each femur and CT acquisition protocol, including element-wise Young's modulus, nodal maximum and minimum principal strains, and load-to-failure. Further details on the FE model construction, material-property assignment, boundary conditions, loading configuration, and simulation procedure are provided in the file `BBCT-hip.docx`. ## Data organisation and format ### CT data The folders `CTdata\Femurs\F_PXXX` and `CTdata\ESP_Phantom\ESP_PXXX` contain the CT data acquired for the two femurs and the ESP phantom, respectively. The corresponding CT acquisition parameters are provided in `CTdata\CTdata_Phantom.csv`. ### Segmentation The folder `Segmentation` contains CT-derived segmentations of both femurs saved as `PXXXR.stl` and `PXXXL.stl` for the right and left femur, respectively. ### FE meshes The folder `mesh\` contains the two FE meshes used in the study, provided in `.cdb` format. The meshes were generated from the segmentations obtained from the P80 CT protocol. ### Young's Modulus The `YoungsModulus` folder contains two files storing the assigned Young’s modulus values for each element of both femurs: E_values_Left.csv' and 'E_values_Right.csv', corresponding to left and right femur, respecitvely. ### Principal strains The folders `PrincipalStrains\Left\` and `PrincipalStrains\Right\` contain the nodal principal strains obtained from the FE simulations for the left and right femurs, respectively. Within each folder, the principal strain data are organised according to the type of strain: - `MaximumPrincipalStrains\` - `MinimumPrincipalStrains\` The files are named according to the following convention: `MaximumStrain_Left_AngleXX.csv` and `MinimumStrain_Left_AngleXX.csv` for the left, and `MaximumStrain_Right_AngleXX.csv` and `MinimumStrain_Right_AngleXX.csv` for right femur, where `XX` identifies the impact pose, which is specified in the file `ImpactPose.csv`. ### Load-to-failure The folder `Load-to-failure\` contains the load-to-failure values obtained from the FE simulations. The files are `Load-to-failure_Left.csv` and `Load-to-failure_Right.csv` for the left and right femur, respectively. Each file contains the load-to-failure values for the corresponding femur, CT acquisition protocol, and impact pose. ## Licensing This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. [Altai2019]: https://doi.org/10.1016/j.clinbiomech.2019.06.004 "Z. Altai, M. Qasim, X. Li, M. Viceconti (2019), Clin Biomech, 68:137-143" [Bhattacharya2019]: https://doi.org/10.1007/s10237-018-1081-0 "P. Bhattacharya, Z. Altai, M. Qasim, and M. Viceconti (2019), Biomech Model Mechanobiol, 18(2):301–318" [Taddei2007]: https://doi.org/10.1016/j.medengphy.2006.10.014 "F. Taddei, E. Schileo, B. Helgason, L. Cristofolini, M. Viceconti (2007), Med Eng Phys, 29(9):973-9" [Aldieri2023]: https://doi.org/10.1016/j.cmpb.2023.107727 "Aldieri A, Curreli C, Szyszko JA, La Mattina AA, Viceconti M (2023), Comput. Methods Programs Biomed. 240:107727" [Aldieri2024]: https://doi.org/10.1038/s41598-024-57618-6 "Aldieri A, Biondi R, La Mattina AA, Szyszko JA, Polizzi S, Dall'Olio D, Curti N, Castellani G, Viceconti M (2024), Sci Rep 14:7403"