Development of an organ‐specific insert phantom generated using a 3D printer for investigations of cardiac computed tomography protocols

Abstract Introduction An ideal organ‐specific insert phantom should be able to simulate the anatomical features with appropriate appearances in the resultant computed tomography (CT) images. This study investigated a 3D printing technology to develop a novel and cost‐effective cardiac insert phantom derived from volumetric CT image datasets of anthropomorphic chest phantom. Methods Cardiac insert volumes were segmented from CT image datasets, derived from an anthropomorphic chest phantom of Lungman N‐01 (Kyoto Kagaku, Japan). These segmented datasets were converted to a virtual 3D‐isosurface of heart‐shaped shell, while two other removable inserts were included using computer‐aided design (CAD) software program. This newly designed cardiac insert phantom was later printed by using a fused deposition modelling (FDM) process via a Creatbot DM Plus 3D printer. Then, several selected filling materials, such as contrast media, oil, water and jelly, were loaded into designated spaces in the 3D‐printed phantom. The 3D‐printed cardiac insert phantom was positioned within the anthropomorphic chest phantom and 30 repeated CT acquisitions performed using a multi‐detector scanner at 120‐kVp tube potential. Attenuation (Hounsfield Unit, HU) values were measured and compared to the image datasets of real‐patient and Catphan® 500 phantom. Results The output of the 3D‐printed cardiac insert phantom was a solid acrylic plastic material, which was strong, light in weight and cost‐effective. HU values of the filling materials were comparable to the image datasets of real‐patient and Catphan® 500 phantom. Conclusions A novel and cost‐effective cardiac insert phantom for anthropomorphic chest phantom was developed using volumetric CT image datasets with a 3D printer. Hence, this suggested the printing methodology could be applied to generate other phantoms for CT imaging studies.


Introduction
Over the past few years, there has been increased use of 3D printing technology for rapid prototyping of highquality printed objects. 1 Since its introduction, 3D printing technology has been successfully applied in numerous areas, such as engineering, industry, art, education and medicine. 2,3 In medicine, 3D printing technology has been used for a variety of purposes, for example, assisting surgical planning, 4 guiding interventional procedures, 5 manufacturing radiology components, 6 printing personalised artificial parts 7 and recently, developing phantoms. [8][9][10][11] Phantoms have been widely applied in medical imaging, especially in CT systems, commonly for both quantitative and qualitative assessments of image quality. Many prior studies [12][13][14] have highlighted the advantages of using phantoms, especially when the investigations involve multiple radiation exposures with different acquisitions settings. One of the most common phantoms used for the investigations of CT protocols is the anthropomorphic chest phantom (Kyoto Kagaku, Japan). This phantom has properties that are very similar to the anatomical features of an adult chest region, for example, the lungs, bones and muscles. However, the cardiac insert of this phantom has single, homogenous material that is not appropriate to simulate CT images, especially for cardiac CT. An ideal cardiac insert should be able at least to simulate the heart features with appropriate appearances in the resultant CT images.
Recently, many recent phantom studies [8][9][10][11]15,16 have employed the 3D printing technology to construct their phantoms. For example, Solomon et al., 16 asserted that this technology could be applied to generate anthropomorphic texture phantoms that are feasible to assess the quality of CT images. Another was by Shuai Leng et al., 15 which used 3D printing technology to generate a comprehensive quality assurance phantom. However, the major drivers are the limitations of available commercial phantoms which are often costly and not customisable. This 3D printing technology allows researchers to design and construct physical phantoms and organ inserts based on their preferences at a lower cost than any commercially available physical phantoms. Additionally, the successful and validated 3D-printed physical phantoms can be reproduced by any other accessible 3D printers. Therefore, it is indeed possible to fabricate 3D-printed phantoms with specific characteristics to suit various imaging investigations, particularly in CT systems. In this study, the investigation of 3D printing technology offers an alternative to produce a novel and cost-effective cardiac insert phantom containing a contrast-enhanced region directly from volumetric CT image datasets of anthropomorphic chest phantom. The printing methodology used in this study could be generally applied to generate other phantoms for CT imaging studies.

Materials and Methods
The following three steps were taken to develop the new cardiac insert phantom: 1 Step one involved obtaining acquisitions of volumetric datasets from a multi-detector CT scanner. 2 Step two involved delineating the regions of interest (ROI) from the surrounding structures, which resulted in segmented image datasets. This step also included the optimisation procedure, such as smoothing and wrapping. 3 Step three involved printing the new physical phantom and removing unnecessary supporting structures so as to produce at final clean physical 3D-printed cardiac insert phantom.
The phantom was printed using a 3D printer of Creatbot DM Plus Model (Mankati, Shanghai, China). This 3D printer uses a fused deposition modeling (FDM) technique to develop the phantom. This FDM technique is similar to inkjet printing but a filament is used instead of ink. Acrylonitrile butadiene styrene (ABS) filament passes through a moving heated extruder to print 0.25 mm layer of material onto the build tray. Next, a cooling fan solidifies the ABS material creating a traced layer onto the tray. The process was then repeated for each layer until completed. Additional support materials were also printed on the layers to prevent the structures from collapsing.
The proceeding sections elaborate on the (1) designs of the 3D-printed cardiac insert phantom, (2) process of printing the physical models, (3) after-printing process and (4) measurement of attenuation (HU) values.

Phantom design
The 3D-printed cardiac insert phantom was made of two main assemblies, (i) the heart-shaped shell and (ii) the removable inserts: (i) The heart-shaped shell design was derived from the volumetric CT image datasets of an anthropomorphic chest phantom (Lungman N-01, Japan; see Fig. 1A-B).
The cardiac insert was scanned on a multi-detector CT scanner (Alexion, Toshiba Medical Systems Co Ltd., Otowara, Japan) using a 120-kVp tube potential and a fixed 200 mA tube current. The reconstructed image datasets were transferred to a segmentation software program (3D Slicer, The Slicer Community, Harvard) 17 to delineate and outline structures from the carina to the apex of the heart phantom, see Figure 1C-D. Next, the segmented image datasets were exported to generate a virtual 3D-isosurface of the cardiac insert using Rhinoceros 3D (McNeel, Seattle, WA, USA) and Autodesk 123D Design (Autodesk Inc., San Rafael, CA, USA) software. To remove any defect, smoothing and surface rendering methods were performed on the 3D-surface mesh. The mesh of the cardiac insert was then saved in a binary stereolithography (STL) file format before it was exported to the 3D printer for printing. The heartshaped shell was designed to ensure it could be suitably positioned in the anthropomorphic chest phantom for CT scanning. of the heart. The diameters of the coronary arteries and the ascending aorta were set from 1.5 to 5.0 mm and 30 mm, respectively. A minimum diameter of 1.5 mm is the smallest diameter detectable as in accordance to the American Heart Association (AHA) Guidelines. 18 The dimensional size of these two removable inserts was further adjusted so that they fitted and could be suitably positioned in the heartshaped shell. Figure 2A-C shows the cross-sectional diagram and the virtual 3D-isosurface of the removable inserts A and B.

Printing process
Three printing tasks had been employed to facilitate the printing process, which were: (i) Insert A; (ii) Insert B; and (iii) Heart-shaped shell. Insert A was printed by segregating it into Parts I and II. Part I refers to the cylindrical structures, while Part II denotes the base layer (Fig. 3A). Meanwhile, Insert B was divided into three parts (Parts I, II and III). Part I refers to the top layer, while Part II denotes the ventricle-shape and Part III reflects the outermost cylinder shape in which to insert both the removable inserts (Fig. 3B). As for the heart-shaped shell, it was separated into Parts I and II, where Part I is for the shell where the removable inserts could be placed, whereas Part II is the top layer (Fig. 3C). Such division of printing parts or assemblies allowed easy filling for the varied density materials, especially after the printing process. The printer settings used during the 3D printing process were configured based on the Simplify3D (Ohio, USA) software program, as shown in Table 1.

After-printing process
Additional support materials, for example, rafts and pillars, were removed from the 3D-printed cardiac insert phantom. Next, the external surface of heart-shaped shell and removable inserts (Inserts A and B) was covered with acrylonitrile butadiene styrene (ABS) liquid to prevent leakage of the materials. This ABS liquid was produced by soaking the ABS filaments into acetone for approximately 30-45 min. All the removable inserts were glued together after the process of filling the phantom with materials of different densities was completed. The heart-shaped shell that supported the two inserts was then filled with jelly to simulate the myocardium. Insert A was filled with oil,    19 Insert B was filled with water material and separately with air material, where the latter simulated the trachea.

Attenuation properties
The average attenuation (Hounsfield Unit, HU) values were measured to verify the properties of the phantom for cardiac CT imaging. All measurements were performed at the CT scanner workstation. The 3Dprinted cardiac insert phantom was positioned in the anthropomorphic chest phantom and imaged 30 times with a multi-detector CT scanner (Alexion, Toshiba Medical Systems). The acquisitions of the phantom were performed at 120-kVp tube potential, scan FOV 250 mm and 0.75-s rotation time. The tube current was set at 200 mA and dose modulation was turned off. The projection image datasets were reconstructed by applying only filtered back projection (FBP) and FC18 reconstruction kernel with a 1.0-mm slice thickness and an axial FOV of 160-mm. The average HU values were measured by placing the ROI over each material (contrast media, air, oil, and jelly) reconstructed axial images of the cardiac phantom, the relevant anatomy (ascending aorta, air, fat and muscle) of patient image datasets and also the air and LDPE inserts of Catphan â 500 phantom (The Phantom Laboratory, Salem NY, USA). Both patient and Catphan â 500 phantom datasets were scanned at similar acquisition protocols.

Results
The physical models and the axial CT images of the completed 3D-printed cardiac insert phantom are illustrated in Figures 4 and 5 respectively. The total printing time was 12.1 hours and phantom preparation time, for example, removing support materials, covering surfaces with ABS liquid, assembling all parts and filling the phantom with materials was 10.2 hours. The cost of the phantom production was approximately US$70, which covered the costs of the ABS filament and the internal materials used. However, the cost of the 3D printer was excluded due to institute ownership.
The mean attenuation (HU) values for circular ROI placed over varied materials within 3D-printed cardiac insert phantom, real-patient image datasets, and Catphan â 500 are tabulated in Table 2 for FBP image reconstruction algorithms. As a result, the measured values confirmed that the materials used in the 3Dprinted cardiac insert phantom are comparable with those obtained from real-patient image datasets and the standard CT image quality phantom Catphan â 500 phantom (Air and LDPE inserts).

Discussion
This paper presents a novel design of a 3D-printed cardiac insert phantom for an anthropomorphic chest phantom, including the associated 3D printing methodology. This phantom was comprised of a contrastenhanced region to enable the investigation of the impact of various settings upon cardiac CT protocols. In a prior work, 20 the use of this new cardiac insert phantom had been demonstrated to determine the impact of various image reconstruction algorithms on image quality and dose reduction potential. The results were consistent with past studies 21-23 as the image datasets reconstructed with iterative reconstruction algorithm exhibited more noise reduction, hence resulting in higher image quality, when compared to the filtered back projection.
To ascertain image quality, researchers 24-27 measured image noise by placing the ROI within a specific anatomical contrast-enhanced region to ascertain image quality. For cardiac CT imaging of coronary CT angiography, the ROI is usually placed within the ascending aorta. [24][25][26] In clinical case, this anatomical region refers to the time-to-peak enhancement of the contrast media, which has been often applied to test the adequacy of the contrast path, and therefore, overall contrast enhancement level as well as diagnostic image quality. 28 For the 3D-printed cardiac insert phantom, a cylindrical contrast-enhanced region was designed with similar diameter to the average ascending aorta (~30 mm). The large size of this cylindrical contrast-enhanced region allowed for the measurement of image noise.
Despite image noise, most clinical-based studies 29-31 also employed the detectability of coronary arteries to determine the subjective image quality that resulted from varying protocols. For instance, Carrascosa et al., 30 determined the overall image quality score based on coronary artery visualisation. As for the present cardiac insert phantom, the varying size of cylindrical contrast-   regions represented this purpose. Hence, the edges and the detectability of these cylindrical contrastenhanced regions over various protocols applied could be used to determine the overall subjective image quality.
Another advantage of this new insert phantom is the removable inserts. This new feature allowed the researchers to further customise the design or the filling materials used to suit their purposes. Additionally, this design was successfully developed by using a CAD software program, hence making it possible for other researchers to redesign and reproduce new physical phantom models. In fact, numerous other open sources software programs are also available on the internet for users to download and use to build their phantom designs.
The primary challenge of 3D printing had been seeking the most apt printing methodology, which is inclusive of selecting suitable printing materials, determining the correct temperature settings of the extruders, and choosing the most appropriate printer protocols. 32 From this work of developing the present phantom, deciding on the appropriate temperature for the extruder to lay the printing materials on the platform had been an intricate issue. Another problem that was experienced had been during the printing process of the removable inserts due to the surface intricacy and the size of subtle diameters.
The new insert phantom offers a good alternative to researchers who need to produce custom phantoms relatively quickly and cost-effectively. Sophisticated phantom production demands the use of the latest 3D printer technology that allows a greater variety of filament materials and the ability to customise phantoms with the desired geometrical features. There are several limitations related to the 3D printer used. First, the new insert phantom resembled a static physical model of a  dynamic organ meaning the various changes that take place during the cardiac cycle were not displayed in the projection images. Different printing materials moved by electrical motors could offer a sequence of mechanical events similar to heartbeats. Second, more advanced 3D printing technology can produce physical models with highly intricate surfaces and sides. This advanced technology could be extended to produce a phantom from the real-patient volumetric CT datasets. This 3D-printed cardiac insert phantom was also comparable with the HU values obtained from the realpatient image datasets and the Catphan â 500 phantom. Hence, it was likely that for all the filling materials, the resulting image quality assessments did display similar results upon using the real-patients or Catphan â 500 phantom. Nonetheless, in any case, additional investigations, for example, resolutions and detectability, using other tools are indeed necessary to ensure that the image quality assessments are accurate.

Conclusions
In conclusion, this study demonstrates that a novel 3Dprinted cardiac insert phantom can be produced from volumetric CT images. This new insert phantom could be reproduced by investigators who need a relatively costand time-effective method of producing customised CT phantoms. Further advances in this 3D printing technology promise to offer more flexibility in design, and this could become a more routine method in producing phantoms in future.