Home > Projects > CranioSim


[bibshow file=http://ressl.fh-linz.at/dokuwiki/doku.php?id=bib:resslpubs,other.bib]

Within this project, the aim is to develop a surgical simulator for cranial grafts harvesting techniques.


The first transplantation of a bone graft was reported in 1668, by the Dutch surgeon Job van Meekeren, who transplanted a graft harvested from a dog's skull into a soldier's cranial bone [bibcite key=DeBoer1988]. Nowadays, bone grafts are harvested from distant sites, like the iliac crest, the tibia, maxilla, mandible, ribs, sternum and from the calvarial bone [bibcite key=Zouhary2010]. Graft harvested from the calvarial bone has become a quite popular site for donoring autologuous grafts, especially for grafts used in cranio-maxillo-facial surgery [bibcite key=Yoshimura1990].

In 1982, the first use of autogenous calvarial bone as a graft material in face reconstruction was reported by Tessier [bibcite key=Tessier1982]. The cranial bone offers numerous advantages as graft material, which include a reduction of post-operative pain of the donor site, a well hidden scar beneath the scalp, a decreased hospitalisation time, and in case of maxilla-facial surgery, a single surgical field. Bone grafts from the skull undergo less resorption and may revascularise more rapidly [bibcite key=Bruno1994].

Due to the reason, that the skull is not flat but convex, the gathering of a passable transplant of the outer table is not that easy. During the harvest, it is absolutely necessary to avoid an entering of the skull's vault [bibcite key=Yoshimura1990]. The disadvantage of this donor site may lie within the risk of a dural tear, contour deformity of the skull and only a minimal amount of cancellous bone which is available for harvest [bibcite key=Zouhary2010].

Figure 1: Preparation for graft extraction (Prim. Malek, AKH Linz). Figure 2: Extracted graft (Prim. Malek, AKH Linz).

The skull vault's bone consists of three layers; the outer and inner table (tabulae externa and interna) consist of compact bone whereas the middle layer, called diploe, consists of spongy, trabecular bone . The preferred harvest site for calvarial grafts is the parietal bone, because of its thickness [bibcite key=Bruno1994].

A typical technique (see Figure 3) to harvest calvarial bone is performed as follows:
after the scalp is cut open and held open with clamps, the area around the area which will be harvested is cut with a cutting burr. The depth of this cut depends on the thickness of the calvarial bone. Then, an oscillating saw can progress below the diploic space. As last step, the final cut is performed with a curved osteotome, which cuts the last diploic connections between the outer and inner table of the cranial bone [bibcite key=Yoshimura1990] in order to obtain the autologuoues grafts (Figure 2).

Figure 3: Procedure for harvesting calvarial bone. A – patient, B – simulator.


Artificial Cranial Bone

In order to set up a simulator for parietal graft lift training, polyurethane (PU)-based bone surrogate composites (Figure 4) enriched with mineral fillers, were developed [bibcite key=Hollensteiner2017d,Hollensteiner2017c,Hollensteiner2017b]. In order to mimic a spongy bone covered by a cortical layer, thus resembling the composition of human skull, four different material compositions were created. To create an open cell cancellous structure, the PU resin was supplemented with water as a blowing agent and a cell stabiliser. This cancellous PU mixture was further supplemented with two fillers- calcium phosphate (CaP, Ca3(PO4)2) and calciumcarbonate (CaC, CaCO3)- in amounts of 50 (CaP050, CaC050) and 100% (CaP100, CaC100) of the whole mass to vary structure and hardness. After curing, the cancellous bone surrogate blocks were cut with a band saw achieving a flat surface which was coated with a second mixture. This cortical mixture, which consisted of the PU resin and the mineral-based fillers, was smoothly spread on the cancellous bone blocks with the aid of a spatula, resulting in an even layer of 1.5 mm. Thus, the artificial cortical layer was in accordance with a human outer table, which thickness is around 1.6 ± 0.2 mm [bibcite key=Jung2003].

Patient Phantom

A training simulator by means of a patient phantom for harvesting cranial bone grafts was built with artificial bone surrogates. The artificial skull cap was placed on a silicone base with a brain-shaped bulge to prevent slipping. Furthermore, the skull cap was covered with a layer of artificial soft tissues (muscles and skin). After a skin incision and the retraction of the soft tissues, the surgeon is able to train the same procedural steps as in a real parietal graft lift surgery. The simulator prototypes were tested by two experienced surgeons (see Figure 5).

Figure 4: Artificial skull cap of the simulator showing cross-sections of artificial bone. Figure 5: Surgeon training on the simulator with real instruments.

The following movie shows a surgeon extracting a cranial graft.


The simulator, represented by the customised patient phantom was validated into two directions: by means of characteristic parameters for drilling, milling and sawing by comparison with human parietal bone and due to an evaluation study gaining face, content and constructive validity measures.

Surgical machining

In accordance with the surgical procedure of a split thickness tabula externa graft lift the maximum puncture forces of the drill- and mill heads penetrating the tabula externa were used to compare the machining characteristics of artificial and human skull bones (see Figure 6). Furthermore, the maximum force of the cancellous sawing procedure, which is due to a full sawblade insertion into the cancellous bone, was identified as a characteristic parameter.

Figure 6: Drilling force over insertion depth at different positions: A: tabula externa, B: diploe, C: tabula interna.


The maximum drilling forces in axial direction during the investigation of the testing blocks were recorded for CaC050 bone surrogates (2.5 ± 0.2 N). In contrast, the recorded forces during the drilling of the human specimens were 1.8 ± 0.5N. The artificial bone materials had 18% (CaP050), 20% (CaP100), 42% (CaC050) and 32% (CaC100) higher average maximum drilling force than the tested human bones. The Shapiro-Wilk test showed normal distribution for all drilling measurement results and homogenous variances were determined for all groups. Significant differences between the human specimens and the bone surrogates were not detected (see Figure 7).


The lowest axial milling forces were detected for surrogate CaC050 (1.3 ± 0.1 N) which were clearly lower than the milling forces observed for human parietal bones (1.7 ± 0.3 N). However, the mean maximum milling forces were 18% (CaP050), 21% (CaP100), 25% (CaC050) and 16% (CaC100) smaller than the results of human specimens. The Shapiro-Wilk test detected a non-normal distribution of the human specimen. Because of this a Mann-Whitney-U-test was used to detect differences between human and artificial specimens. All bone surrogates revealed high p-values (see Figure 7).


The Shapiro-Wilk test indicated normal distribution for all sawing measurements and the Levene-test showed homogenous variances. Thus, an unpaired Student‘s -t test was used to detect differences between human and artificial bones. The the lowest sawing forces were found for CaC050 (0.5 ± 0.1 N). Thus, the mean maximum sawing forces of the artificial bones CaP050, CaP100, CaC050, CaC100 were 32%, 11%, 43%, 26% smaller than the human bone samples (see Figure 7).

Figure 7: Milling-, drilling- and sawing forces for artificial materials in comparison to human parietal bone.

Anatomical layers

To create an artificial skull cap with anatomically realistic bone layers, the bone layer thickness values of the tabula externa, diploe and tabula interna as well as the total thickness of the parietal bone samples were investigated (see Figure 8). The results were used as a reference for the development of the bone layers of the artificial skull cap (ASC), especially for the estimation of the amounts of materials needed to mold the bone layers. All human specimens were scanned with a micro-computed tomography (CT80, Scanco 41 Medical, Brüttisellen, Switzerland). For all images, resolution preferences of 70 kV, 114A and a 200ms integration time were used resulting in a slice thickness of 0.09mm.
For all samples, the thickness of the inner and outer cortical layers as well as the total thickness were measured. Hence, the thickness of the diploic space could be calculated by subtracting the cortical layer thickness values from the total bone thickness.

Figure 8: A – cross section of a human parietal bone sample, B – artificial bone. ET/IT – tabula externa/interna thickness, DT – diploe thickness, TT – total thickness.

The results are summarised by Figure 9. Human bone samples showed a total bone thickness of 5.31 ± 1.24mm. The total thickness of the artificial skull cap (ASC) was 6.94 ± 1.20mm and thus were 30.1% larger than the human total thickness. The Sawbones (SB) skull samples had a total thickness of 12.51 ± 1.14mm and though were more than twice as thick (+135.4 %) as the human bone samples. The externa thickness values of the human bones were 1.26 ± 0.34mm. The artificial skull caps were similar (1.15 ± 0.25mm) and only were 9 % larger than the human tabula externa while the SB samples were more than 165 % larger (3.35 ± 0.73mm).

Similar results were found for the interna thickness values. The average interna thickness of the human parietal bone was 0.76 ± 0.22mm. ASC (1.07 ± 0.34mm, +40.7%) and SB (3.69 ± 1.18mm, +385.5%) were clearly thicker.

Figure 9: Boxplots for interna/externa thickness, diploe thickness and total thickness for the developed artificial skull cap (ASC), commercially available artificial bones from Sawbones (SB) as well as human samples.

Validation study

The aim of this study was to verify the face, content and construct validity of the novel model-based simulator within a validation study [bibcite key=Hollensteiner2017e]. Participants were enforced to perform a tabula externa graft lift during a training session on the simulator. Task performance was analysed with a standardised assessment tool.
Prior to the training task, the participants (n=15) completed questionnaires based on demographics and their medical experiences. Then, experts and novices were invited to perform lifts of a 2×1.5cm graft split in three equal proportions from the parietal tabula externa using the model-based simulator.
The study-participants rated the realism and appropriateness as a teaching modality of the TabSIM along a 5-point Likert scale. The realism, or face validity, was confirmed with an average score of 4.2 of a maximum of 5 points. Likewise, the content validity, which proved the appropriateness of the simulator as a teaching modality, was proven with an average score of 4.8 of maximum 5 points. During the evaluation of task performance, the expert surgeons were evaluated significantly better than the novices, which proved the construct validity of the simulator (see Figure 10).

Figure 10: Boxplots showing the differences between novices and experts for the global rating scale, the transplant area and the surgery time.