A 65-year-old Asian male presented to our centre with a history of lung cancer with brain metastasis. He was treated with radiotherapy in 2002, after which he developed sinonasal carcinoma. He underwent cranionasal resection and craniectomy in February 2020 by an ear, nose and throat surgeon and neurosurgeon following neoadjuvant chemoradiation. Recovery was complicated by cerebrospinal fluid (CSF) leakage from the nose. An attempt was made to repair the defect with temporalis fascia and lumbar drainage of the CSF, however, the leakage persisted. A second repair attempt was made in June 2020 using a fascial graft and a free vastus lateralis muscle flap. Because the superficial temporal vessels had been sacrificed in the previous surgery, the pedicle vessels had to be grafted and anastomosed to vessels in the neck. A 21 cm radial artery graft and a long saphenous vein graft were harvested and anastomosed in end-to-end fashion using interrupted nylon suture. The pedicle artery was anastomosed to the radial artery graft, then to the right facial artery. The pedicle vein was anastomosed to the long saphenous vein graft, then to the internal jugular vein. The pedicle vessel ran through a subcutaneous tunnel along the patient’s right temple and preauricular area. The frontal bone was not returned after craniectomy in order to avoid strangulating the muscle flap. The patient recovered uneventfully afterwards.
The resultant frontal skull defect from the supraorbital rim to the vertex measured 64 mm vertically and 95 mm across (Fig. 1). An external helmet was necessary to prevent injury to this region of the brain during rehabilitation. However, our conventional ready-made helmets were deemed unsuitable for two reasons: First, the anterior edge of the conventional helmet landed 1 cm above the eyebrow line and would have violated the craniectomy defect (Fig. 2); second, the conventional helmet was fixed by straps across the temple and submental regions bilaterally, which would have compressed the vein grafts and neck vessel anastomoses. To circumvent these issues, a perfectly fitting, strapless 3D-printed helmet was designed.
Computed tomography (CT) scans of the head were obtained in Digital Imaging and Communications in Medicine (DICOM) format with 1 mm slice thickness and a 512 × 512 region of interest (ROI) resolution, corresponding to a voxel size of 0.410 × 0.410 × 1 mm. The images were imported into Mimics v21 software (Materialise NV, Leuven, Belgium) and segmented using the low-noise soft tissue convolution kernel and Hounsfield unit thresholding (Fig. 3). A two-layered soft tissue and bony 3D model was created in Standard Triangle Language (STL) format and refined using Meshmixer v3.5 software (Autodesk, San Rafael, CA, USA) and 3-matic v13 software (Materialise). Mesh reduction (decimation), smoothing, and defect filling techniques were applied to the 3D head model to reduce its triangular complexity. A smoothed layer with a 2 mm offset from the skin surface was created to accommodate the patient’s hair (Fig. 4). The model was further smoothed at the craniectomy site to prevent impingement by the helmet and abrasion at the edge of the site.
The 3D head model was printed for helmet design and liner fabrication (Fig. 5a). The helmet was designed to rest on the supraorbital bar and wrap around the vertex and occipital area to avoid compressing the craniectomy defect. A 6 mm layer of Plastazote® foam was fitted to the model and shaped such that the pressure-sensitive areas were avoided (Fig. 5b). A window was created over the right temple and facial area in order to avoid compression of the free vastus lateralis muscle flap pedicle. The foam-fitted head model was scanned using a Spectra 3D scanner (Vorum, Vancouver, Canada). The scanned model was imported to Canfit computer-aided design (CAD) software (Vorum), which was used to define the borders of the helmet and produce a 6 mm-thick prototype helmet model (Figs. 6 and 7).
The helmet model was then 3D printed using medically designated (ISO 10993 and USP Class VI biocompatibility certified) acrylonitrile butadiene styrene (ABS-M30i, Stratasys, MI, USA) by an industrial-grade fused deposition modelling (FDM) printer (Fortus 450mc, Stratasys). The model was printed with a layer thickness of 0.178 mm with a sparse infill density set at 18% to decrease weight and improve patient comfort. The helmet was lined with the foam insert and tested for fitting on the 3D head model (Fig. 8).
Upon wearing, the helmet was confirmed to fit properly, without the need for straps to secure it to the patient’s head (Fig. 9). The patient had a prolonged stay in a rehabilitation hospital for physiotherapy. Four months postoperatively, he has undergone three months of daily physiotherapy, including focused lower limb strength training in the form of cycling and kicking exercises. He can now walk with fair stability using a rollator and the assistance of another person. His Glasgow Coma Scale (GCS) score is E4V2M6 and he remains dependent on others for self-care and activities of daily living. The helmet has proven sufficiently durable and remains wholly intact and in use at the time of writing.