**3. 3D image acquisition**

3D facial data can be obtained using a variety of different techniques. CT and MRI scans have been used for the RP of internal body organs and structures as well as facial prosthesis prototyping. These techniques have an advantage over external body 3D scanning as they can be used for generating patient specific 3D models of internal body parts or organs, etc. However, due to their lower resolution, they are unable to capture fine textural details like pores and wrinkles on the surface of the body. Furthermore, data capture is undertaken in supine positions and thus can affect the accuracy of captured data in that the positional detail is not truly reflective of that in a natural upright repost position. Finally, there is no ability to capture colour data using these modalities.

Compared to CT and MRI, 3D photogrammetry is becoming more popular due to the expe‐ diency of the data capture, lack of exposure to ionising radiation, ability to capture data in a prone position, and relatively easy storage, transfer and utilisation of the data. Furthermore, the cost of such systems is relatively affordable. Active 3D scanning techniques like laser scanning [21], Kinect [22], structured light [23] and speckle projection stereo [24] have also been used previously to obtain 3D scanned data for manufacturing of facial prosthetics. Although the image acquisition and 3D reconstruction techniques are different using these methods, the outputs produced are similar in terms of the data produced and they all produce 3D mesh data and colour information for the scanned geometry.

Typically surface scanners, including the 3dMD scanner, are based on the speckle projection technique and was used successfully in this project. This system is based on stereo photo‐ grammetry and calculates stereo correspondence by projecting infrared speckle patterns over the target surface to then create a 3D depth map of the scanned surface. It consists of several machine vision cameras and infrared speckle projectors combined with a flash system to acquire texture and 3D topographic information simultaneously. These are collectively known as pods. The process then merges all the viewpoint data from each pod to create the final 3D polygon mesh with the colour images then laid over the 3D mesh to add colour, texture and appearance. The resulting mesh can then be saved as an .OBJ file for processing and editing purposes. An example of this scanned data is shown in **Figure 2**.

As with the surface topographic information, the colour images from the 3dMD camera system may also require further processing before they can be overlaid on the 3D mesh prior to final colour printing. This primarily involves colour management of the 2D colour image from the camera RGB to printer RGB for each pixel using specific camera and printer colour profiles respectively. For this specific task, the colour profiles used were developed using both conventional colorimetric and spectral‐based reproduction methods [19, 20]. When the colour management is finalised, surface texture mapping is conducted to map the newly generated colour image onto the manipulated 3D model. The penultimate step involves 3D colour printing to produce the 3D model using the 3D printing system, and then, finally as a post‐ processing step, the strength and flexibility of the printed model are improved by infiltrating

3D facial data can be obtained using a variety of different techniques. CT and MRI scans have been used for the RP of internal body organs and structures as well as facial prosthesis prototyping. These techniques have an advantage over external body 3D scanning as they can be used for generating patient specific 3D models of internal body parts or organs, etc. However, due to their lower resolution, they are unable to capture fine textural details like pores and wrinkles on the surface of the body. Furthermore, data capture is undertaken in supine positions and thus can affect the accuracy of captured data in that the positional detail is not truly reflective of that in a natural upright repost position. Finally, there is no ability to

Compared to CT and MRI, 3D photogrammetry is becoming more popular due to the expe‐ diency of the data capture, lack of exposure to ionising radiation, ability to capture data in a prone position, and relatively easy storage, transfer and utilisation of the data. Furthermore, the cost of such systems is relatively affordable. Active 3D scanning techniques like laser scanning [21], Kinect [22], structured light [23] and speckle projection stereo [24] have also been used previously to obtain 3D scanned data for manufacturing of facial prosthetics. Although the image acquisition and 3D reconstruction techniques are different using these methods, the outputs produced are similar in terms of the data produced and they all produce

Typically surface scanners, including the 3dMD scanner, are based on the speckle projection technique and was used successfully in this project. This system is based on stereo photo‐ grammetry and calculates stereo correspondence by projecting infrared speckle patterns over the target surface to then create a 3D depth map of the scanned surface. It consists of several machine vision cameras and infrared speckle projectors combined with a flash system to acquire texture and 3D topographic information simultaneously. These are collectively known as pods. The process then merges all the viewpoint data from each pod to create the final 3D polygon mesh with the colour images then laid over the 3D mesh to add colour, texture and appearance. The resulting mesh can then be saved as an .OBJ file for processing and editing

it with medical grade silicone.

92 New Trends in 3D Printing

**3. 3D image acquisition**

capture colour data using these modalities.

3D mesh data and colour information for the scanned geometry.

purposes. An example of this scanned data is shown in **Figure 2**.

**Figure 2.** (a)–(c). a) Speckle projection images and texture bitmap captured from a single pod. (b) 3D mesh created 3 using all viewpoint data. (c) The 2D bitmap image overlaid over 3D mesh to add colour.

To ensure sufficient quality images, protocols can be developed to aid data capture. These can include—capturing images in a blackout facility, minimising patient movement, employ consistent peripheral lighting protocols and following the camera manufacturers recommended protocol for image capture, including geometric and colour calibrations immediately prior to imaging. When capturing images to use in the 3D printing of facial prostheses, patients should be asked to remain motionless during the image capturing process in order to avoid any motion artefacts. If the patient has already had surgery to remove the affected facial part, they will often have precision attachments retaining existing prostheses. In cases such as these, due to the limitations of data capture for these attachment—small componentry, shiny surfaces, obscured by anatomical features including prominences, undercuts and recesses, acrylic bosses (described below) can be used to ensure the camera captures the exact location of any precision attachments being used. Ideally, the patient should also be maintained a neutral pose during imaging.
