Desktop 3D printing in medicine to improve surgical navigation in acral tumors

Abstract

Purpose: Image guidance may improve limb-sparing surgery results. For this purpose, preoperative computed tomography (CT) or magnetic resonance (MR) images can be registered to patient’s anatomy by means of a tracking system. In this approach, a rigid transformation is commonly applied for the registration step. This assumption may not be correct for acral tumors because distal extremities have many joints with complex movements (e.g. hand has 27 degrees of freedom). For reducing target registration error (TRE), a similar limb position must be ensured in the preoperative images and during image guided surgery (IGS). Additive manufacturing and rapid prototyping are easily available for clinical use thanks to three-dimensional (3D) printing. In orthopedic surgery, patient-specific anatomical models have been created for several purposes such as preoperative planning, pre-bending plates design, training and patient-physician communication. Patient-specific 3D surgical aids are used to improve the surgical planning and assist the surgeon during the procedure [1]. Regarding desktop 3D printing applied to orthopedic surgery, good results have recently been documented in reconstructive pelvic surgery [2] and in techniques of lower-limb realignment [3]. In this context, we propose to take one step further using a desktop 3D printer to design and create patient-specific distal extremity molds that ensure a similar position during imaging and IGS in the operating room (OR). The aim of this abstract is to describe a new workflow that merges IGS and desktop 3D printing in limb sparing surgery of distal extremities. Our study also evaluates the reproducibility of distal extremity position during navigation using a patient-specific 3D-printed hand phantom. Methods: The first step in the suggested workflow for IGS in acral tumors starts by printing a 3D mold of the distal extremity. This holder will lay on the surgical bed allowing the limb to remain in a fixed and known position. The mold cannot be too tight since pressure may cause edema to the patient. The holder is modeled by extruding a patient’s hand surface created from the segmentation of a previous CT image. Alternatively, structured-light 3D scanners could be used in this step, thus avoiding ionization radiation. Several conical holes (Ø 4 mm x 3 mm depth) are modeled on the mold surface to be used as landmarks for the pointer in the registration step. Moreover, three screws (designed to attach optical passive markers) are included in the layout to define a reference frame that would allow accounting for mold movements during navigation. The modeling step is done with freely available software (MeshMixer and 123D Design, Autodesk, Inc., USA). The desktop 3D printer used is Witbox-2 (BQ, Spain), a low-cost fused deposition modeling (FDM) hardware managed with open-source software. The thermoplastic material chosen is polylactic acid (PLA) because of its extrudability and nontoxic properties. Furthermore, the resulting holder can be sterilized by ethylene oxide [4]. The next step involves the surgical planning (segmentation of tumor and definition of surgical margins) using the radiological postprocessing software Horos (GNU open-source OsiriX,www.horosproject.org) on previous CT or any CT/MR study acquired with the limb placed on the printed mold. Conical holes can be used to facilitate the registration between CT and MR images. Finally, during IGS, the distal extremity is placed on the sterilized mold. The navigation is performed with an optical tracking system (OTS) after registering the conical holes of the holder in the image space with those corresponding ones obtained with a tracked pointer in the OR (physical space). A patient with a soft-tissue sarcoma in the palm of his right hand was selected for evaluating the reproducibility of the limb placement on its mold during navigation. The holder (Fig. 1) was created from the CT used to plan the neoadjuvant external radiotherapy. A 3D model of the hand was also printed separately from the mold. This mold (Fig. 1) included nine conical holes (Ø 4 mm x 3 mm depth) on its surface for placing the pointer tip in the validation step. To resemble the patient’s CT image, a CT scan was acquired with the rigid hand on its mold (voxel size 0.5 x 0.5 x 0.5 mm). Navigation was performed with a multi-camera OTS (OptiTrack, NaturalPoint Inc., USA) with 7 cameras, which reduces occlusion problems caused by OR personnel and surgical support devices and has previously been evaluated [5]. The tracking system was connected by means of Plus Toolkit (www.assembla.com/spaces/plus/wiki) to a 3D Slicer platform (www.slicer.org) with a SlicerIGT extension. The evaluation consisted in placing the 3D printed hand on its mold, registering the conical holes (mold) of the CT (image space) with those corresponding ones obtained with a tracked pointer (physical space) and, finally, estimating the TRE between the conical holes (hand) in the image space and in the physical space. These steps were repeated four times. Results: Table 1 shows the fiducial registration error (FRE) obtained from the conical holes in the mold and TRE from the conical holes in the printed hand. Figure 2 displays the physical space and the image space after registration. Conclusion: This study presents an IGS workflow for acral tumors that includes desktop 3D printing for reproducing distal extremity position. A multidisciplinary team of surgeons and engineers worked together in the process of modeling and printing the patient-specific mold with a low-cost FDM printer at the hospital. TRE was in accordance with a previous study using the same tracking system [5], demonstrating the reproducibility of hand position during navigation. These results allow us to follow this procedure during the final intervention that is scheduled in one month’s time. Results with real patient data will be presented during the conference.