Introduction
This document describes an example of using Piper framework to personalize and position the human body model GHBMC M50-OS to a PMHS subject (PMHS711) selected among a series of PMHS pedestrian tests. Pedestrian finite element models (PFEM) are used to investigate and predict the injury outcomes from vehicle-pedestrian impact. Comparison of PFEM mechanical response in full-scale scenarios using Post-Mortem Human Surrogates (PMHS) is mandatory to ensure the biofidelity of such models. As postmortem human surrogates (PMHS) differ in anthropometry across subjects, the biofidelity of PFEM cannot be properly evaluated by comparing a generic anthropometry model against the specific PMHS test data [Paas 2015]. Geometrical personalization of PFEM using morphing was shown to increase the biofidelity of PFEM [Poulard et al. 2016]. In addition, similar benefit to the evaluation was observed when reproducing the correct posture of the model [Paas 2015]. Since the PIPER software framework includes tools for both positioning and the personalization of HBM, it is expected that the framework will facilitate the overall process. The purpose of the document was to illustrate the benefit of the Piper tools for positioning and personalization of HBM for in an in-depth example that could represent real scenarios of use. Consequently, this document was written so it can be easily implemented as a tutorial for advanced users of Piper.
What files are needed for this tutorial?
This tutorial assumed that the user have access to the following:
- A version of GHBMC M50-PS
- Piper metadata associated to GHBMC M50-PS (“.pmr) that will describe the component of the model for Piper (see Help on this topic)
- Simplified Scalable Model linking anthropometric dimensions defined in ANSUR database to GHBMC M50-PS entities (“.xml”)
- Piper framework (available here)
Note: as Piper was designed to be model and code agnostic, this tutorial could be applied to any PFEM assuming that the user have access to the necessary Piper metadata files.
Approach
This document will described how to 1) morph a PFEM to a specific PMHS using the anthropometric measurements available on the PMHS and to 2) position the PFEM using the coordinates of a series of landmarks on the body surface acquired on the PMHS prior to impact.
Step 1. Morphing using the Anthropometry Module
Preparing the data
Table 1 shows the main anthropometric dimensions of measured on the PMHS identified as PMHS711. Measurements were taken according to Rebiffé, Guillien, Pasquet (Anthropometrical inquiry on the French drivers, 1981-1982).
Table 1. Anthropometric dimensions of PMHS711 (original).
| Label | Value (mm) |
|---|---|
| Age | 78 y.o |
| Gender | Male |
| Weight | 64 |
| Stature | 168 |
| Pelvic breadth (iliac crest) | 265 |
| Bitrochanteric breadth | 320 |
| Hip circumference | 760 |
| Head circumference | 580 |
| Head breadth | 150 |
| Head depth | 185 |
| Chest height | 870 |
| Acromion-seat height | 640 |
| Chest depth (axilla) | 165 |
| Chest breadth (axilla) | 275 |
| Chest circumference (axilia) | 900 |
| Chest depth (xiphoid process) | 220 |
| Chest breadth (xiphoid process) | 290 |
| Chest circumference (xiphoid process) | 920 |
| Arm length | 345 |
| Forearm length | 265 |
| Forearm + hand length | 450 |
| Upper leg length | 490 |
| Lower leg length | 520 |
| Shoulder breadth | 380 |
Alone, this data may not be enough to perform the scaling of the HBM. Consequently, this data will be completed with additional anthropometric dimensions that will be predicted using the Anthropometry Module (see the associated tutorial for more details). The Anthropometry Module relied on the statistical process detailed in Parkinson and Reed 2010 which used the ANSUR database to generate anthropometric targets. The anthropometric dimensions will be then used in the module Scaling Constraint Module to scale the model.
The ANSUR database used 636 anthropometric measurements (also called predictors) that could differed from the source of data for morphing. Consequently, the user should identify those, among the ANSUR predictors, correspond to the anthropometric measurements presented in the desired source (Table 2). Definitions of the ANSUR predictors can be found online or directly in the Piper directory: (\PIPER\share\octave\anthropoPerso\dataBase\Dataset_Ansur_1989\DataReference\ANSUR_OriginalData). In this example, age was excluded from the predictor as the target age (78) was not in the range of the ANSUR database (17-50) and could lead to regression issues.
Table 2. Anthropometric predictors of PMHS711 (ANSUR).
| ANSUR Predictor | Value |
|---|---|
| WEIGHT | 64 |
| STATURE | 1680 |
| HIP_BRTH_SITTING | 320 |
| WAIST_CIRC-OMPHALION | 760 |
| HEAD_CIRC | 580 |
| HEAD_BRTH | 150 |
| HEAD_LNTH | 185 |
| SITTING_HT | 870 |
| ACR_HT-SIT | 640 |
| CHEST_DEPTH | 165 |
| CHEST_BRTH | 275 |
| CHEST_CIRC | 900 |
| ELBOW_REST_HT | 295 |
| ACR_RADL_LNTH | 265 |
| FOREARM-HAND_LENGTH | 450 |
| KNEE_HT_-_SITTING | 520 |
| BIDELTOID_BRTH | 380 |
Generate anthropometric dimension by regression using Piper
This section describes how to generate target files of coherent anthropometric dimensions by using the Anthropometry Module and use them to scale a Human Body Model.
First, let’s setting up the anthropometry module
- Run Piper
- Open the Anthropometry Module
- Click on "From dataset" in the right menu.
- Click on "New regression". A new subsection will appear: "DatasetOptions".
- Click on "Select Dataset" button, and then select the "Adult Dataset(ANSUR)"
- Click on “Select Variables”. Now we select the weight from predictors listed below by typing WEIGHT in the search field and then left-click on it.
-
Repeat the previous step for the following predictors: STATURE, HIP_BRTH_SITTING, WAIST_CIRC-OMPHALION, HEAD_CIRC, HEAD_BRTH, HEAD_LNTH, SITTING_HT, ACR_HT-SIT, CHEST_DEPTH, CHEST_BRTH, CHEST_CIRC, ELBOW_REST_HT, ACR_RADL_LNTH, FOREARM-HAND_LENGTH, KNEE_HT_-_SITTING, BIDELTOID_BRTH
-
Then click one the "Set Population Descriptors" subsection, select one bin, representing 100% of the population, with the gender as "male", and set the age from 30 to 51 years.
- Click on "Define target information" subsection. Select a "sample type" - choose "MeanOnly". S elect a "sample input type" - choose "Fixed predictor". Leave "sample output type" as "Normal
- A new subsection called "Set Predictor Values" appears, that will allow you to set values for the previously selected predictors. Note that the order of appearance of the predictors in this window is directly linked to the order of selection of the predictors in “Select Variables”.
- In Output Options, Click on "Set Target File path", and define the location you want to save the target file in, in the right menu.
- Once your regression and sample information is defined, all the traffic lights should be green, and now you can click on the "Generate Targets" button. It will create the target file in the specified directory. It takes usually less than a minute to generate the file.
Once you have generated some target files with the anthropometry module, you are able to use them across piper modules to personalize your model.
Personalize geometrically the model using Piper
- Open the GHBMC M50-OS model by importing a new pmr model (see Project menu for details on importing files). Depending on your configuration, it can take up to a minute.
- Once loaded, select the Scaling Constraint module
- Import the target file you generated in the Define samples information step using the Target menu.
- Click on the "Scalable Model" button and then on "Import Simplified Scalable Model" and select xml file with the description of the Simplified Scalable Model for ANSUR.
- Click on the "Target Dimensions" button and then on "Use current targets". The simplified scalable model and associated control points are visualized in white color, the target in blue.
- Click on Perform Deformation. Tick on “Display intermediate skin/bone target” to visualize a preview of bone and/or skin deformed shape. If satisfied, click on “Apply- with intermediate target”
- The scaling is now completed. It is adviced to save the piper session so that it can retrieve later. To do so, click on “Project” then “Save as” and locate the appropriate folder where the session will be saved.
Step 2. Positioning
Preparing the data
The position of the subject was quantified by recording the coordinates of a series of landmarks on the body surface prior to impact (Table 3). This set of data will be used to interactively position the model using the coordinates of the landmarks in Table 3 as target.
Table 3. Positioning measurements for PMHS711 (target).
| Anatomical description | X (mm) | Y (mm) | Z (mm) |
|---|---|---|---|
| Heel_R | -98 | 113 | 687 |
| Tiptoe_R | 175 | 147 | 671 |
| Heel_L | 245 | -128 | 671 |
| Tiptoe_L | 537 | -118 | 662 |
| Ankle_External_Malleolus_R | -34 | 156 | 605 |
| Ankle_Internal_Malleolus_R | -7 | 87 | 604 |
| Ankle_External_Malleolus_L | 308 | -173 | 597 |
| Ankle_Internal_Malleolus_L | 320 | -109 | 592 |
| Fibula_Styloid_Process_R | 89 | 183 | 268 |
| Fibula_Styloid_Process_L | 231 | -180 | 249 |
| High_Pubic_Symphysis | 199 | 0 | -173 |
| Great_Trochanter_L | 148 | -162 | -176 |
| Great_Trochanter_R | 150 | 178 | -176 |
| Ulna_Styloid_Process_L | 329 | -17 | -202 |
| Ulna_Styloid_Process_L | 358 | 39 | -208 |
| Anterior_superior_iliac_spine_R | 195 | 108 | -250 |
| Anterior_superior_iliac_spine_L | 197 | -86 | -268 |
| Humerus_Lateral_Epicondyle_R | 260 | 179 | -371 |
| Humerus_Lateral_Epicondyle_L | 254 | -164 | -386 |
| Sternum_Sternal_Notch | 238 | 19 | -685 |
| Angulus_Acromialis_R | 217 | 227 | -698 |
| Angulus_Acromialis_L | 192 | -176 | -706 |
| Ear_Canal_R | 228 | 98 | -830 |
| Ear_Canal_L | 244 | -57 | -839 |
| Nose_Nasal_Root | 342 | 34 | -883 |
In this example, the coordinates are already expressed in the same coordinate system as the model. Similarly, the user has to locate the same landmarks on the HBM (after scaling). The user has to ensure that both sources of data ( Table 3 and Table 4) shared the system frame as the model (origin and orientation). It is recommended to use the landmark closest to the pelvis as origin to optimize the results.
Table 4. Positioning measurements for GHBMC M50-OS before positioning and after scaling
| Anatomical description | Nodes | X | Y | Z |
|---|---|---|---|---|
| Heel_R | 901383 | -130 | 89 | 757 |
| Tiptoe_R | 901029 | 164 | 110 | 730 |
| Heel_L | 701314 | 184 | -98 | 754 |
| Tiptoe_L | 702026 | 478 | -112 | 726 |
| Ankle_External_Malleolus_R | 9019600 | -80 | 124 | 647 |
| Ankle_Internal_Malleolus_R | 9000608 | -47 | 65 | 646 |
| Ankle_External_Malleolus_L | 7034365 | 241 | -126 | 655 |
| Ankle_Internal_Malleolus_L | 7053103 | 275 | -70 | 638 |
| Fibula_Styloid_Process_R | 9091818 | 38 | 140 | 272 |
| Fibula_Styloid_Process_L | 7072071 | 173 | -142 | 274 |
| High_Pubic_Symphysis | 8074854 | 199 | 0 | -173 |
| Great_Trochanter_L | 8067743 | 91 | -170 | -172 |
| Great_Trochanter_R | 8068847 | 82 | 170 | -174 |
| Ulna_Styloid_Process_L | 3002271 | 288 | -68 | -252 |
| Ulna_Styloid_Process_R | 5000542 | 300 | 56 | -291 |
| Anterior_superior_iliac_spine_R | 8068008 | 157 | 135 | -241 |
| Anterior_superior_iliac_spine_L | 8067415 | 159 | -134 | -241 |
| Humerus_Lateral_Epicondyle_R | 5004047 | 126 | 252 | -397 |
| Humerus_Lateral_Epicondyle_L | 3004185 | 126 | -251 | -397 |
| Sternum_Sternal_Notch | 4001959 | 167 | 0 | -713 |
| Angulus_Acromialis_R | 4000074 | 87 | 185 | -726 |
| Angulus_Acromialis_L | 4000024 | 75 | -188 | -719 |
| Ear_Canal_R | 1043059 | 89 | 81 | -902 |
| Ear_Canal_L | 1024994 | 89 | -81 | -902 |
| Nose_Nasal_Root | 1007995 | 210 | 0 | -917 |
Positioning interactively the model using Piper
- In the Positioning module, click on “Pre”. The positioning module will start loading. Note that the initialization of the module can take up to 10 min.
- In the Control parameters (Control), set bone collision off. Deactivating this option decrease the time of computation by a significant factor and will allow a real time interaction when positioning.
-
To position model, it is recommended to position the model by regions (left arm, head, leg, foot). In this example we will reposition both lower extremities using the landmarks located on the foot (Ankle_External_Malleolus_L and Ankle_External_Malleolus _R). To avoid the whole model to move while positioning, we will isolated the upper body.
- To fix the upper body, click on “Fixed bones” and select “Head”, “Body_proper”, Right_upper_limb” and Left_upper_limb”. The fixed regions will be highlighted in red.
- To add the target landmarks of the foot, click on Landmark. Select Ankle_External_Mallelous_L, tick “X” and “Y” and click on “+”. Repeat the operation for Ankle_External_Mallelous_R.
- You will notice for each landmarks, 2 sets of coordinates: one with colors (red and green) represents your objective while the set in black will be the current positions. Based on Table 3: click on the colored set and type the values to reach (see figures below).
- Click on Positioning and observe the positioning of the model in real time. When satisfied click again on Positioning.
- Repeat the operations for the left arm, the head and the right arm.
- When satisfied, update the model nodes coordinates
- Finally, you can export the positioned model back to the FE code format in the Project menu
