Session Item

Thermoplastic masks are currently the standard in RT immobilization. These can however be uncomfortable due to shrinkage, induce claustrophobia and can limit beam angle selection in treatment plans. We therefore propose a novel concept for more comfortable patient-specific and plan-optimized head and neck immobilization devices using the full potential of additive manufacturing (AM) and Computer-Aided Design (CAD). 

To design the immobilization, CT data of a head and neck patient was retrospectively collected. The body contour was segmented from the head up to the shoulders in Mimics® (Materialise) and exported as a 3D mesh. All subsequent CAD was performed in fusion360® (Autodesk). First, the body mesh was converted into a solid modelling boundary representation and positioned on a concept model of the envisioned treatment table, including two attachment sites for the immobilization. Next, a two-part mask was designed around the patient’s body. The lower shell includes the shoulders and a cranial stop and the upper part the nose bridge and the chin (figure 1). Using AI assisted generative design (GD) the connection between the lower shell and the table attachments were generated using a downward load case of 35 N to represent an adult human head and HP PA12 nylon as the printing material of choice. Regions to be avoided such as a treatment machine bore or nozzle can be taken into account in the GD process. A Finite Element Analysis (FEA) static stress test was performed to calculate the factor of safety (FoS) and expected displacement. 

The proposed workflow allowed for the creation of a patient-specific open face mask starting from CT data. We were able to integrate additional functionalities to the immobilization structure such as a beam modifier (bolus/range shifter) and in-vivo dosimetry which can ensure accurate placement of these devices with respect to the patient anatomy. The immobilization device could also be customized to the treatment plan, by avoiding the beam having to pass through material or structural edges which can disturb the intended dose distribution in the patient (figure 1). This can ultimately allow more flexibility in the beam angle selection, potentially resulting in an improved treatment plan quality. The static stress test indicated a realized FoS of 9.6, indicating that further optimisation of the support could reduce the use of material and production cost given that a safety factor of 2-3 should suffice. A maximal displacement of 0.16 mm was indicated (figure 2).

The realised FoS and maximal displacement given by the FEA indicates that our novel immobilization using GD and AM is a viable concept. We therefore conclude that the proposed workflow shows to be a promising new concept for patient immobilization. Future work would include investigating the use of surface scanning and extensive testing of the stability, reproducibility and comfortability of the immobilization devices using phantoms and eventually patients.