research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775

Low-density tissue scaffold imaging by synchrotron radiation propagation-based imaging computed tomography with helical acquisition mode

crossmark logo

aDivision of Biomedical Engineering, College of Engineering, University of Saskatchewan, Saskatoon, SK S7N 5A9, Canada, bDepartment of Anatomy, Physiology and Pharmacology, College of Medicine, University of Saskatchewan, Saskatoon, SK S7N 5A9, Canada, cDepartment of Mechanical Engineering, College of Engineering, University of Saskatchewan, Saskatoon, SK S7N 5A9, Canada, dDepartment of Chemical and Biological Engineering, College of Engineering, University of Saskatchewan, Saskatoon, SK S7N 5A9, Canada, and eCanadian Light Source, Saskatoon, SK S7N 2V3, Canada
*Correspondence e-mail: xbc719@usask.ca, niz504@usask.ca

Edited by A. Stevenson, Australian Synchrotron, Australia (Received 28 April 2022; accepted 29 January 2023; online 16 February 2023)

Visualization of low-density tissue scaffolds made from hydro­gels is important yet challenging in tissue engineering and regenerative medicine (TERM). For this, synchrotron radiation propagation-based imaging computed tomography (SR-PBI-CT) has great potential, but is limited due to the ring artifacts commonly observed in SR-PBI-CT images. To address this issue, this study focuses on the integration of SR-PBI-CT and helical acquisition mode (i.e. SR-PBI-HCT) to visualize hydro­gel scaffolds. The influence of key imaging parameters on the image quality of hydro­gel scaffolds was investigated, including the helical pitch (p), photon energy (E) and the number of acquisition projections per rotation/revolution (Np), and, on this basis, those parameters were optimized to improve image quality and to reduce noise level and artifacts. The results illustrate that SR-PBI-HCT imaging shows impressive advantages in avoiding ring artifacts with p = 1.5, E = 30 keV and Np = 500 for the visualization of hydro­gel scaffolds in vitro. Furthermore, the results also demonstrate that hydro­gel scaffolds can be visualized using SR-PBI-HCT with good contrast while at a low radiation dose, i.e. 342 mGy (voxel size of 26 µm, suitable for in vivo imaging). This paper presents a systematic study on hydro­gel scaffold imaging using SR-PBI-HCT and the results reveal that SR-PBI-HCT is a powerful tool for visualizing and characterizing low-density scaffolds with a high image quality in vitro. This work represents a significant advance toward the non-invasive in vivo visualization and characterization of hydro­gel scaffolds at a suitable radiation dose.

1. Introduction

In scaffold-based tissue engineering and regenerative medicine (TERM), scaffolds made from biomaterials are used to support and facilitate cell growth and tissue regeneration, as well as transport nutrients and metabolic wastes (Chen, 2019[Chen, D. X. (2019). Extrusion Bioprinting of Scaffolds for Tissue Engineering Applications. Cham: Springer.]). For this, scaffolds should possess appropriate architectural, mechanical and biological properties to mimic those of native tissues or organs. As such, visualization and/or characterization of scaffold properties via in vitro and/or in vivo studies are essential to TERM applications (Duan et al., 2021[Duan, X., Li, N., Chen, X. & Zhu, N. (2021). Tissue Eng. C, 27, 573-588.]). In scaffold-based TERM, hydro­gels are the most commonly used biomaterials for scaffolds with water-swollen crosslinked polymeric networks (Nezhad-Mokhtari et al., 2019[Nezhad-Mokhtari, P., Ghorbani, M., Roshangar, L. & Soleimani Rad, J. (2019). Eur. Polym. J. 117, 64-76.]). Typically, hydro­gels have similar density to the surrounding environment (e.g. water for in vitro study or soft tissue after implantation for in vivo study).

To visualize and characterize hydro­gel scaffolds in TERM, synchrotron radiation computed tomography (SR-CT) imaging holds promise for both in vitro and in vivo applications (Duan et al., 2021[Duan, X., Li, N., Chen, X. & Zhu, N. (2021). Tissue Eng. C, 27, 573-588.]; Zhu et al., 2011[Zhu, N., Chapman, D., Cooper, D., Schreyer, D. J. & Chen, X. (2011). Tissue Eng. C, 17, 1071-1080.]; Ning et al., 2018[Ning, L., Sun, H., Lelong, T., Guilloteau, R., Zhu, N., Schreyer, D. J. & Chen, X. (2018). Biofabrication, 10, 035014.], 2021[Ning, L., Zhu, N., Smith, A., Rajaram, A., Hou, H., Srinivasan, S., Mohabatpour, F., He, L., Mclnnes, A., Serpooshan, V., Papagerakis, P. & Chen, X. (2021). Appl. Mater. Interfaces, 13, 25611-25623.]; Olubamiji et al., 2014[Olubamiji, A. D., Izadifar, Z. & Chen, D. X. (2014). Tissue Eng. B, 20, 503-522.]; Naghieh & Chen, 2021[Naghieh, S. & Chen, X. (2021). J. Pharm. Anal. 11, 564-579.]; Izadifar, Babyn, Kelly et al., 2017[Izadifar, M., Babyn, P., Kelly, M. E., Chapman, D. & Chen, X. (2017). Tissue Eng. C, 23, 548-564.]; Izadifar, Babyn, Chapman et al., 2017[Izadifar, M., Babyn, P., Chapman, D., Kelly, M. E. & Chen, X. (2017). J. Synchrotron Rad. 24, 842-853.]; Olubamiji et al., 2016[Olubamiji, A. D., Izadifar, Z., Zhu, N., Chang, T., Chen, X. & Eames, B. F. (2016). J. Synchrotron Rad. 23, 802-812.], 2017[Olubamiji, A. D., Zhu, N., Chang, T., Nwankwo, C. K., Izadifar, Z., Honaramooz, A., Chen, X. & Eames, B. F. (2017). Tissue Eng. C, 23, 156-168.]; You et al., 2016[You, F., Wu, X., Zhu, N., Lei, M., Eames, B. F. & Chen, X. (2016). ACS Biomater. Sci. Eng. 2, 1200-1210.]; Bawolin & Chen, 2016[Bawolin, N. K. & Chen, X. (2016). J. Funct. Biomater. 7, 17.]). For in vitro scaffold visualization, SR-CT is non-destructive/non-invasive imaging which means scaffolds can be imaged without the need for sectioning and other processes. As such, it allows accurate capture of scaffold structures in longitudinal studies of the same scaffolds. It is noted that, among various in vitro imaging techniques, scanning electron microscopy (SEM) (Bartoš et al., 2018[Bartoš, M., Suchý, T. & Foltán, R. (2018). Biomed Eng. OnLine, 17, 110.]; Stachewicz et al., 2019[Stachewicz, U., Szewczyk, P. K., Kruk, A., Barber, A. H. & Czyrska-Filemonowicz, A. (2019). Mater. Sci. Eng. C, 95, 397-408.]; Vitas et al., 2019[Vitas, S., Segmehl, J. S., Burgert, I. & Cabane, E. (2019). Materials, 12, 416.]) has been commonly used for morphology and pore/surface structure of the scaffolds with the wide magnification range 10–500 000× (Zhu et al., 2021[Zhu, X., Hu, Y., Wu, G., Chen, W. & Bao, N. (2021). ACS Nano, 15, 9273-9298.]), but the approach involves destructive sample preparation. Such destructive processes may change the structure (e.g. pore size) of scaffolds, leading to inaccurate results and findings. Confocal laser scanning microscopy (CLSM) (Bagherzadeh et al., 2013[Bagherzadeh, R., Latifi, M., Najar, S. S., Tehran, M. A. & Kong, L. (2013). J. Biomed. Mater. Res. 101A, 765-774.]; Phipps et al., 2012[Phipps, M. C., Clem, W. C., Grunda, J. M., Clines, G. A. & Bellis, S. L. (2012). Biomaterials, 33, 524-534.]) is another commonly used technique for tissue scaffold visualization, which enables collection of data in 3D but is limited to the penetration depth range 200–300 µm (Parrilli et al., 2014[Parrilli, A., Pagani, S., Maltarello, M., Santi, S., Salerno, A., Netti, P., Giardino, R., Rimondini, L. & Fini, M. (2014). J. Microsc. 255, 20-29.]). CLSM has been commonly used for quantitative biological analysis and can provide complementary information on tissue scaffolds to SR-CT imaging (morphology). For in vivo scaffold visualization, SR-CT has the merits of high spatial resolution and high contrast, as well as a relatively fast scan speed. Magnetic resonance imaging (MRI) (Mueller et al., 2021[Mueller, K. M. A., Topping, G., Schwaminger, S. P., Zou, Y., Rojas-González, D. M., De-Juan-Pardo, E. M., Berensmeier, S., Schilling, F. & Mela, P. (2021). Biomater. Sci. 9, 4607-4612.]; Chen et al., 2020[Chen, J., Cai, Z., Wei, Q., Wang, D., Wu, J., Tan, Y., Lu, J. & Ai, H. (2020). J. Mater. Chem. B, 8, 316-331.]; Kotecha et al., 2017[Kotecha, M., Magin, R. L. & Mao, J. J. (2017). Magnetic Resonance Imaging in Tissue Engineering. Hoboken: Wiley.]), ultrasound imaging (UI), photoacoustic (PA) imaging (Teodori et al., 2017[Teodori, L., Crupi, A., Costa, A., Diaspro, A., Melzer, S. & Tarnok, A. (2017). J. Biophoton, 10, 24-45.]) and optical coherence tomography (OCT) (Chen et al., 2011[Chen, C.-W., Betz, M. W., Fisher, J. P., Paek, A. & Chen, Y. (2011). Tissue Eng. C, 17, 101-112.]; Wang et al., 2018[Wang, L., Xu, M., Luo, L., Zhou, Y. & Si, P. (2018). Sci. Rep. 8, 2802.]) can visualize scaffolds non-destructively and non-invasively. However, MRI requires a long scan time to achieve a high spatial resolution [e.g. about 100 h for 100 µm (Edlow et al., 2019[Edlow, B. L., Mareyam, A., Horn, A., Polimeni, J. R., Witzel, T., Tisdall, M. D., Augustinack, J. C., Stockmann, J. P., Diamond, B. R., Stevens, A., Tirrell, L. S., Folkerth, R. D., Wald, L. L., Fischl, B. & van der Kouwe, A. (2019). Sci. Data, 6, 244.])], whereas SR-CT needs only seconds to minutes to achieve a much higher resolution [e.g. lower than 10 µm at dose rates suitable for in vivo applications (Harrison et al., 2022[Harrison, K. D., Sales, E., Hiebert, B. D., Panahifar, A., Zhu, N., Arnason, T., Swekla, K. J., Pivonka, P., Chapman, L. D. & Cooper, D. M. (2022). J. Bone Miner. Res. 37, 2244-2258.])]. Hydro­gel imaging with a long-time scan may change some of the properties of a hydro­gel, leading to motion artifacts. High-resolution imaging (<100 µm) of UI/PA requires high-frequency scanning, which may heat up scanned scaffolds. Besides, for in vivo UI, the presence of bone can greatly limit the resolution. OCT can achieve a high resolution, but its penetration depth is a shortcoming compared with SR-CT, particularly for full scaffold imaging in vivo in animal models (e.g. rats). Compared with conventional absorption contrast CT imaging using a conventional (polychromatic) X-ray tube as a source, monochromatic SR-CT can obtain both absorption contrast and phase contrast. The latter can achieve a higher contrast for low-density scaffolds, especially when combined with phase retrieval (PhR). It is noted that SR-CT is a broad class of techniques that includes interferometer CT (Miao et al., 2016[Miao, H., Panna, A., Gomella, A. A., Bennett, E. E., Znati, S., Chen, L. & Wen, H. (2016). Nat. Phys. 12, 830-834.]), edge-illumination CT (Hagen et al., 2014[Hagen, C., Diemoz, P., Endrizzi, M., Rigon, L., Dreossi, D., Arfelli, F., Lopez, F., Longo, R. & Olivo, A. (2014). Opt. Express, 22, 7989-8000.]; Momose et al., 2003[Momose, A., Kawamoto, S., Koyama, I., Hamaishi, Y., Takai, K. & Suzuki, Y. (2003). Jpn. J. Appl. Phys. 42, L866-L868.]), diffraction-enhanced imaging CT [SR-DEI-CT (Chapman et al., 1997[Chapman, D., Thomlinson, W., Johnston, R., Washburn, D., Pisano, E., Gmür, N., Zhong, Z., Menk, R., Arfelli, F. & Sayers, D. (1997). Phys. Med. Biol. 42, 2015-2025.])], analyzer-based imaging [SR-ABI-CT (Wernick et al., 2003[Wernick, M. N., Wirjadi, O., Chapman, D., Zhong, Z., Galatsanos, N. P., Yang, Y., Brankov, J. G., Oltulu, O., Anastasio, M. A. & Muehleman, C. (2003). Phys. Med. Biol. 48, 3875-3895.])] and propagation-based imaging CT [SR-PBI-CT (Suzuki et al., 2002[Suzuki, Y., Yagi, N. & Uesugi, K. (2002). J. Synchrotron Rad. 9, 160-165.])]. SR-PBI-CT has the great advantage of simple implementation and fast acquisition compared with SR-DEI-CT/SR-ABI-CT (Izadifar et al., 2016[Izadifar, Z., Honaramooz, A., Wiebe, S., Belev, G., Chen, X. & Chapman, D. (2016). Biomaterials, 82, 151-167.]). As such, it has great potential for clinical studies (Fedon et al., 2018[Fedon, C., Rigon, L., Arfelli, F., Dreossi, D., Quai, E., Tonutti, M., Tromba, G., Cova, M. A. & Longo, R. (2018). J. Med. Imaging, 5, 013503.]; Longo et al., 2014[Longo, R., Tonutti, M., Rigon, L., Arfelli, F., Dreossi, D., Quai, E., Zanconati, F., Castelli, E., Tromba, G. & Cova, M. A. (2014). Philos. Trans. R. Soc. A, 372, 20130025.]; Castelli et al., 2011[Castelli, E., Tonutti, M., Arfelli, F., Longo, R., Quaia, E., Rigon, L., Sanabor, D., Zanconati, F., Dreossi, D., Abrami, A., Quai, E., Bregant, P., Casarin, K., Chenda, V., Menk, R. H., Rokvic, T., Vascotto, A., Tromba, G. & Cova, M. A. (2011). Radiology, 259, 684-694.]) and in vivo animal imaging (Taba et al., 2018[Taba, S. T., Gureyev, T. E., Alakhras, M., Lewis, S., Lockie, D. & Brennan, P. C. (2018). Am. J. Roentgenol. 211, 133-145.]). Therefore, this work involved the application of SR-PBI-CT to hydro­gel scaffold imaging.

Despite many advantages, SR-PBI-CT image quality for low-density scaffolds still requires improvement. Ring artifacts, for example, are an issue for SR-PBI-CT (Pelt & Parkinson, 2018[Pelt, D. M. & Parkinson, D. Y. (2018). Meas. Sci. Technol. 29, 034002.]) due to systematic errors or defects on the scintillator, monochromator or filters. Such artifacts usually make it difficult to process and analyze images using existing methods for visualizing and charactering samples. This is a particular concern for imaging low-density scaffolds due to their relatively low contrast. Methods have been developed to reduce ring-like artifacts, including pre-processing and post-processing algorithms, but these approaches suffer from various limitations in practical application. For example, low-pass Fourier filtering (Raven, 1998[Raven, C. (1998). Rev. Sci. Instrum. 69, 2978-2980.]) poses the risk of introducing additional artifacts in the reconstructed background. The Sarepy sorting wind method (Vo et al., 2018[Vo, N. T., Atwood, R. C. & Drakopoulos, M. (2018). Opt. Express, 26, 28396-28412.]) usually has limited ring artifact removal efficacy for noisy images and for images with unresponsive stripes which result from dead pixels on the detector and/or damaged areas of the scintillator. In addition, these algorithms involve manual hyperparameter selection (e.g. the size of convolution kernel or window size which are defined by the user). Different values will produce different outcomes and it usually takes a long time to determine the optimal value. Besides, for different regions of interest (ROIs), the hyperparameter values need to be finely tuned for optimal artifact removal and therefore subjectivity is a limitation. These limitations raise a great need to address the issue of ring artifacts if SR-PBI-CT is to realize its larger potential for imaging of low-density targets, including hydro­gel scaffolds.

This paper explores the integration of SR-PBI-CT with helical acquisition mode (hereafter SR-PBI-HCT) to address the ring artifact issues which negatively impact hydro­gel scaffold visualization. Helical acquisition mode has been widely used for clinical CT, but is still relatively novel when combined with SR-PBI-CT. To the best of our knowledge, the SR-PBI-HCT technique, though reported previously (Pelt & Parkinson, 2018[Pelt, D. M. & Parkinson, D. Y. (2018). Meas. Sci. Technol. 29, 034002.]), has not yet been used in the visualization and characterization of scaffolds for tissue engineering applications. To optimize image quality of hydro­gel scaffolds with SR-PBI-HCT, we studied the impact of helical pitch (p, definition given in Section 2.2[link]), X-ray photon energy (E) and the number of acquisition projections per revolution (360°) (Np). Although some of these have been discussed previously (Oliva et al., 2020[Oliva, P., Di Trapani, V., Arfelli, F., Brombal, L., Donato, S., Golosio, B., Longo, R., Mettivier, G., Rigon, L., Taibi, A., Tromba, G., Zanconati, F. & Delogu, P. (2020). Sci. Rep. 10, 17430.]; Taba et al., 2019[Tavakoli Taba, S., Baran, P., Lewis, S., Heard, R., Pacile, S., Nesterets, Y. I., Mayo, S. C., Dullin, C., Dreossi, D., Arfelli, F., Thompson, D., McCormack, M., Alakhras, M., Brun, F., Pinamonti, M., Nickson, C., Hall, C., Zanconati, F., Lockie, D., Quiney, H. M., Tromba, G., Gureyev, T. E. & Brennan, P. C. (2019). Acad. Radiol. 26, e79-e89.]; Nesterets et al., 2018[Nesterets, Y. I., Gureyev, T. E. & Dimmock, M. R. (2018). J. Phys. D Appl. Phys. 51, 115402.]) for regular SR-PBI-CT imaging, optimal parameters for SR-PBI-HCT remain unclear. Furthermore, we evaluated the image quality of hydro­gel scaffolds using SR-PBI-HCT at a low-level radiation dose (i.e. 342 mGy, which is acceptable for in vivo imaging). Our present study, though carried out in vitro, is thus intended to serve as a step towards 3D non-invasive in vivo characterization for TERM applications using hydro­gel scaffolds.

2. Materials and methods

2.1. Scaffold preparation

In this work, 4% w/v aqueous alginate made from medium-viscosity alginate powder (alginic acid sodium salt from brown algae, CAS 9005–38-3, Sigma–Aldrich) and mixed material solutions (3% w/v alginate with 1% w/v gelatin, made from gelatin powder from porcine skin, G1890, Sigma) were prepared. The preparation process was similar to the procedure developed in our previous work (Ning et al., 2021[Ning, L., Zhu, N., Smith, A., Rajaram, A., Hou, H., Srinivasan, S., Mohabatpour, F., He, L., Mclnnes, A., Serpooshan, V., Papagerakis, P. & Chen, X. (2021). Appl. Mater. Interfaces, 13, 25611-25623.]). The prepared solutions were then magnetically stirred at room temperature overnight, or until thoroughly mixed. The crosslinking solution was a calcium chloride dihydrate (CaCl2·2H2O, CAS 10035–04–8, Sigma–Aldrich) solution at 50 mM concentration, with 0.1% w/v polyethyl­ene­imine (PEI, J61270, Alfa Aesar) solution used as the solvent. The surfaces of 12-well plates were coated with 3 ml of the same PEI solution and were left in an incubator at 37°C overnight. The PEI coating was used to ensure that the scaffolds did not stick to the well plate once printing and crosslinking were complete.

The next day, the PEI solution coated on printed plates was replaced with 3–4 ml of the crosslinking solution. The syringes containing different solutions were loaded into the bioprinter and attached to the printing arm. An envisionTEC 4th Generation 3D-Bioplotter Manufacturer Series – an extrusion based (pneumatic) bioprinter – was utilized. Scaffolds were fabricated with dimensions 10 mm × 10 mm × 5 mm with a strand diameter of 500 µm following a grid pattern [Fig. 1[link](a)] with inter-strand distances of 1 mm and 1.5 mm. These were kept in containers in a fridge at 4°C for 2 days.

[Figure 1]
Figure 1
Alginate scaffold: (a) designed scaffold structure and (b) printed scaffolds.

2.2. Synchrotron imaging system setup

The SR-PBI-HCT imaging experiments were performed at the Biomedical Imaging and Therapy (BMIT) 05ID-2 beamline (Wysokinski et al., 2007[Wysokinski, T. W., Chapman, D., Adams, G., Renier, M., Suortti, P. & Thomlinson, W. (2007). Nucl. Instrum. Methods Phys. Res. A, 582, 73-76.]) of the Canadian Light Source. On this beamline, the double-crystal Si(111) monochromator can produce photon energies of 25–140 keV. Fig. 2[link] displays a schematic diagram illustrating the SR-PBI-HCT and SR-PBI-CT imaging setups. The essence of PBI is propagation with distance [i.e. distance from sample to detector (SDD)], which can turn phase distortions into interference fringes and produce large contrast values at the edges of structures. All scans were performed at SDD = 1.5 m, a distance that can provide satisfactory contrast and spatial resolution for hydro­gel scaffold imaging with a reconstructed voxel size of 13 µm based on a previous study by our group (Section S1 of the supporting information). For parallel beam geometry, SR-PBI-HCT scanning usually involves a horizontal rotation range over 360° for achieving the full imaging of the ROI (Section S2 of the supporting information), whereing SR-PBI-CT scanning takes place with a rotation of exactly 180°. The adjustment of photon flux can be achieved through placing neutral density filters (NDFs) with particular thickness in the beam. The detector, with a pixel size of 13 µm and an image depth of 16 bit, consists of a beamline monitor (AA60 HAMAMATSU, Japan) with a scintillator [LuAg500, lutetium Lu3Al5O12 garnet (LuAG) doped by the luminescent Ce3+, thickness of 500 µm] converting X-rays to visible light, an optic system and a complementary metal–oxide semiconductor (CMOS) digital camera (ORCA Flash 4.0). The active areas of the detector for SR-PBI-CT and SR-PBI-HCT imaging are 716 × 2048 pixels (9.31 mm × 26.624 mm) and 200 × 2048 pixels (2.6 mm × 26.624 mm), respectively.

[Figure 2]
Figure 2
Schematic of the SR-PBI-HCT and SR-PBI-CT imaging setups.

As discussed, p is a parameter for SR-PBI-HCT imaging, which can be defined as

[p \,=\, {{{s}_{\rm v}\,t}\over{{h}_{\rm FOV }}}, \eqno(1)]

where sv is the vertical speed (mm s−1) of the rotation stage, t is the required time (s) for a full rotation/revolution and hFOV is the height of the field of view (FOV) (mm) (i.e. the height of the active detector).

In addition, the radiation dose was measured using a calibrated ionization chamber (PTW 31010, Freiburg, Germany). The dose rate [\dot D] (Gy s−1) was measured and then the surface entry radiation dose D (Gy) using SR-PBI-HCT was calculated by

[D_{\rm{SR\hbox{-}PBI\hbox{-}HCT}} \,=\, \dot D \,{{{N}_{\rm p}}\over{p}}\, \Delta t, \eqno(2)]

where Δt is the exposure time (s) for each projection and Np/p is the effective projection number per pitch. Since SR-PBI-CT only involves half a revolution (i.e. 180°) and there is no pitch, the effective projection number is given by Np/2. Given the fact that the beam flux may not be exactly uniform (especially for SR-PBI-CT, as seen later in Fig. 5) and that the surface entry dose rate was measured at a position in front of the geometric center of the detector in our study, the dose rate measured is approximately its maximum value and the dose evaluated from equation (2)[link] is the approximate maximum surface entry dose.

2.3. Imaging of hydro­gel scaffolds with SR-PBI-HCT

Table 1[link] displays the detailed experimental imaging conditions for five different test groups. The image quality of hydro­gel scaffolds using both SR-PBI-HCT and SR-PBI-CT was first evaluated. In addition, in order to determine the spatial resolution of SR-PBI-HCT and SR-PBI-CT, a 3D bar pattern phantom (5 mm × 5 mm, QRM, Möhrendorf, Germany) was also imaged. Then, we conducted three parameter test experiments using SR-PBI-HCT, i.e. varied p (1.3, 1.5, 1.7, 1.9 and 2.5), E (30 keV, 40 keV and 50 keV) and Np (500, 1000 and 3000), and examined their corresponding effects on hydro­gel scaffold image quality qualitatively and quantitatively. Due to the limited active detector height, multiple rotations (i.e. three rotations for p = 1.3, 1.5, 1.7, 1.9, and two rotations for p = 2.5) were taken for imaging the scaffolds when using SR-PBI-HCT. In the end, we scanned the hydro­gel scaffold using SR-PBI-HCT at a low-level radiation dose (i.e. 342 mGy) to evaluate the feasibility of SR-PBI-HCT for in vivo imaging.

Table 1
Experimental details in terms of imaging conditions for five different group tests

Group Scan mode Imaged object Energy (keV) Projection number (rev−1) Binning mode Exposure time (ms per projection) Helical pitch Measured dose (Gy)
No. 1: SR-PBI-CT versus SR-PBI-HCT SR-PBI-CT Hydro­gel scaffold (4% w/v alginate) 30 3000 1 × 1 57.00 79.51
3D QRM bar pattern phantom
SR-PBI-HCT Hydro­gel scaffold (4% w/v alginate) 30 3000 1 × 1 57.00 1.3 122.33
3D QRM bar pattern phantom
No. 2: Helical pitch comparison SR-PBI-HCT Hydro­gel scaffold (3% w/v alginate with 1% w/v gelatin) 30 3000 1 × 1 57.00 1.3 122.33
1.5 106.02
1.7 93.54
1.9 83.70
2.5 63.61
No. 3: Photon energy comparison SR-PBI-HCT Hydro­gel scaffolds (4% w/v alginate) 30 3000 1 × 1 57.00 1.5 106.02
40 14.40 48.38
50 10.82 31.16
No. 4: Projection number per revolution comparison SR-PBI-HCT Hydro­gel scaffold (3% w/v alginate with 1% w/v gelatin) 30 500 1 × 1 57.00 1.5 17.67
1000 35.34
3000 106.02
No. 5: Low-dose imaging SR-PBI-HCT Hydro­gel scaffold (3% w/v alginate with 1% w/v gelatin) 30 250 2 × 2 25.00 1.5 0.34

2.4. Image processing and evaluation metrics

In order to apply existing image reconstruction and post-processing algorithms tailored to SR-PBI-CT, the projections obtained using SR-PBI-HCT needed to be converted to virtual projections acquired by SR-PBI-CT (Pelt & Parkinson, 2018[Pelt, D. M. & Parkinson, D. Y. (2018). Meas. Sci. Technol. 29, 034002.]; Fu et al., 2014[Fu, J., Willner, M., Chen, L., Tan, R., Achterhold, K., Bech, M., Herzen, J., Kunka, D., Mohr, J. & Pfeiffer, F. (2014). Phys. Med. 30, 374-379.]). The steps to form virtual projections in our study are presented schematically in Fig. 3[link] and the Python script is available on GitHub: https://github.com/Xiaoman896/HCT2CT.

[Figure 3]
Figure 3
Schematic of virtual projection stitching of SR-PBI-CT from SR-PBI-HCT projections. Ii,j,k and [{\hat I_{i^{\,\prime},\,j^{\,\prime},\,k^{\,\prime}}}] are the gray value of the SR-PBI-HCT projection and the gray value of the converted virtual SR-PBI-CT projection at the positions (i, j, k) and (i′, j′, k′), respectively. i and i′ are the row indexes, j and j′ are the column indexes, and k and k′ are the projection indexes. [{\hat I_{i^{\,\prime},\,j^{\,\prime},\,k^{\,\prime}}}] is the linear interpolation result of Ii, j, k and Ii+1,j,k with a weight of 1 − w and w. The virtual SR-PBI-CT projection at an index of k′ is transformed from a sequence of SR-PBI-HCT projections at index of k, k + Np/2,…, k + (2NR − 1)Np/2 with an interval of Np/2. The total row M of the SR-PBI-HCT projection is [\lfloor h_{\rm FOV}/d \rfloor] while the total row M′ of the virtual SR-PBI-CT projection is [\lfloor N_{\rm R}(s_{\rm v}/t)+ (h_{\rm FOV}/d)\rfloor]. mod(·,·) denotes the modulo operator.

Assuming the rotation stage is moved downwards, this conversion process can be expressed by

[{\hat{I\,}}_{{i^{\,\prime}}, \, {j^{\,\prime}}, \, {k^{\,\prime}}} = (1-w)\,{I}_{i, \, j,\, k}+w {I}_{i+1, \, j, \, k}, \eqno (3)]

where i and i′ are the row indexes, j and j′ are the column indexes, and k and k′ are projection indexes. Ii, j, k and [\hat I_{i^{\,\prime},j^{\,\prime},k^{\,\prime}}] denote the gray value of the SR-PBI-HCT projection and the gray value of the converted virtual SR-PBI-CT projection at the (i, j, k) and (i′, j′, k′) positions, respectively; w is the linear interpolation weight.

Values i and i′ are i = 1, 2,…, M, where [M = \lfloor h_{\rm FOV }/d \rfloor] and i′ = 1, 2,…, M′, and M′ = [\lfloor N_{\rm R} (s_{\rm v}\,t /d) + (h_{\rm FOV}/d) \rfloor]. NR is the number of rotations and d is the pixel size [sv, t and hFOV have the same definitions as in equation (1)[link]]. Let [v = s_{\rm v}\,t / N_{\rm p}d] be the vertical movement speed (units of pixels per projection), then the vertical translation for the kth projection will be kv (unit of pixels). The relationship between i′ and i can be expressed by

[i^{\,\prime} = i + \lfloor kv \rfloor. \eqno (4)]

Given the fact that there is no displacement/translation in the horizonal direction for SR-PBI-HCT scanning, we have

[\eqalignno{ & j^{\,\prime} \,\mathop =\limits^{\triangle} \ \, \left \{ {\matrix { j,\hfill & {\rm if \ mod}\left(\left \lfloor {{k}/{(N_{\rm p}/2)}}\right \rfloor, 2 \right ) = 0, \cr N-j+1,\hfill & {\rm if \ mod}\left(\left \lfloor {{k}/{(N_{\rm p}/2)}}\right \rfloor, 2 \right ) = 1,}}\right. \cr& {\rm for}\ \ j = 1, 2, \ldots, N, & (5) }]

where mod(…) denotes the modulo operator and N is the total column number of the SR-PBI-HCT projection. The above equation indicates that the column index needs to be horizontally flipped if [\lfloor k/(N_{\rm p}/2) \rfloor] is odd.

k and k′ are the projection indexes and k = 0, 1,…, NRNp − 1 (corresponding angle range from 0 to 360NR°) while k′ = 0, 1,…, Np/2 − 1 (corresponding angle range from 0 to 180°). The relationship between k and k′ is given by

[k^{\,\prime} = {\rm mod} \left (k, {{{N}_{\rm p}}\over 2} \right ). \eqno (6)]

w in equation (3)[link] is the linear interpolation weight and can be calculated by

[w \,=\, kv - \lfloor kv \rfloor. \eqno (7)]

In the present study, we noticed that, if p < 2, there were some redundant data (i.e. same rows appearing in different projections) for the SR-PBI-HCT projections during conversion, which were discarded.

Before the projection conversion from SR-PBI-HCT to virtual SR-PBI-HCT, the background (i.e. flat and dark) corrections were first completed. Then, the transport of intensity equation (TIE) (Paganin et al., 2002[Paganin, D., Mayo, S. C., Gureyev, T. E., Miller, P. R. & Wilkins, S. W. (2002). J. Microsc. 206, 33-40.]), a popular PhR algorithm, was performed on each projection. The δ/β value (2000 for all cases) can be calculated approximately (Thompson & Vaughan, 2001[Thompson, A. C. & Vaughan, D. (2001). X-ray Data Booklet. Lawrence Berkeley National Laboratory, University of California Berkeley, CA, USA.]). The open-source software package Ultra-Fast-Online (UFO) was used to perform PhR (i.e. TIE) on the projections and the CT reconstruction [filtered-backprojection (FBP) algorithm] (Vogelgesang et al., 2016[Vogelgesang, M., Farago, T., Morgeneyer, T. F., Helfen, L., dos Santos Rolo, T., Myagotin, A. & Baumbach, T. (2016). J. Synchrotron Rad. 23, 1254-1263.]). In addition, after PhR, images obtained using SR-PBI-CT were also processed with common ring artifact removal methods, i.e. low-pass Fourier filtering (Raven, 1998[Raven, C. (1998). Rev. Sci. Instrum. 69, 2978-2980.]) and the Sarepy sorting wind method (Vo et al., 2018[Vo, N. T., Atwood, R. C. & Drakopoulos, M. (2018). Opt. Express, 26, 28396-28412.]). For low-pass Fourier filtering, the essence of ring artifact removal is to filter the vertical stripes (i.e. ring artifact areas) in the 2D frequency domain, and the key parameters that need to be tuned are the horizontal and vertical sigma, i.e. 10 and 1 in our work, respectively. For the Sarepy sorting wind method, the essence of ring artifact removal is to employ a median filter (along the horizontal direction) to remove vertical stripes, where the first step is to retrieve the response of each pixel by sorting intensities along each column of a sinogram. The key parameters to be tuned are the window size and the signal-to-noise ratio (SNR). The window size indicates the median filter size and, in our study, a value of 10 (unit of pixels) was selected. The SNR parameter controls the sensitivity of the stripe detection and a value of SNR from 1.1 to 3.0 is recommended. To compare image quality across the different imaging conditions, both objective estimation of quantitative evaluation metrics and subjective evaluation (e.g. image perception and cognition) were employed. Quantitative evaluation metrics include SNR and contrast-to-noise ratio (CNR) (Yao et al., 2019[Yao, L., Zeng, D., Chen, G., Liao, Y., Li, S., Zhang, Y., Wang, Y., Tao, X., Niu, S., Lv, Q., Bian, Z., Ma, J. & Huang, J. (2019). Phys. Med. Biol. 64, 035018.]), which are formulated as

[{\rm SNR} = \left| {{{{I_{{E_{\rm foreground}}}}} \over {{\sigma _{\rm background}}}}} \right|, \eqno (8)]

[{\rm CNR} = \left| {{{{I_{{E_{\rm foreground}}}} - {I_{{E_{\rm background}}}}} \over {{\sigma _{\rm background}}}}} \right|. \eqno (9)]

IE and σ denote the mean gray value and the standard deviation of the ROI, respectively. The foreground represents the ROI including objective hydro­gel scaffolds and the background represents the ROI excluding objective samples.

In addition to SNR and CNR, which mainly focus on the image noise level, the modulation transfer function (MTF) (Fujita et al., 1992[Fujita, H., Tsai, D.-Y., Itoh, T., Doi, K., Morishita, J., Ueda, K. & Ohtsuka, A. (1992). IEEE Trans. Med. Imaging, 11, 34-39.]) of the 3D QRM bar pattern phantom CT images and Fourier shell correlation (FSC) (Van Heel & Schatz, 2005[Heel, M. van & Schatz, M. (2005). J. Struct. Biol. 151, 250-262.]) were also calculated for the spatial resolution analysis. FSC is used to estimate the correlation coefficient between Fourier shells of two 3D reconstructions computed from two independent datasets. In our study, two mutually independent datasets were generated by scanning scaffolds with double projection numbers per rotation and using the odd projections for one reconstruction and the even projections for a second reconstruction.

3. Results and discussion

3.1. Comparison of SR-PBI-HCT and SR-PBI-CT

The reconstructed hydro­gel scaffold (4% w/v alginate) images using SR-PBI-HCT and SR-PBI-CT are shown in Fig. 4[link]. From the results [Figs. 4[link](a1)–4[link](d1)], the SR-PBI-HCT image does not show obvious artifacts whereas the SR-PBI-CT image presents serious ring artifacts. Such artifacts disrupt the continuity of the strand grayscale values in the image and thus impair the capacity to accurately visualize/analyze strand properties. Although these artifacts can be removed somewhat using image-processing algorithms, there are still some obvious remaining artifacts. The gray value profiles across strands [Figs. 4[link](a2)–4[link](d2)], as indicated by the red lines in Figs. 4[link](a1)–4[link](d1), were also visualized, providing the quantitative contrast information.

[Figure 4]
Figure 4
Comparison of SR-PBI-HCT and SR-PBI-CT (gray scale: 0–255); (a1) SR-PBI-HCT image, (b1) SR-PBI-CT image, (c1) SR-PBI-CT image with low-pass Fourier filtering and (d1) SR-PBI-CT image with Sarepy sorting. (a2)–(d2) corresponding gray value profiles of ROIs at red line positions.

SR-PBI-HCT can cause helical artifacts due to the nature of its helical acquisition. Compared with ring artifacts, helical artifacts spread over larger regions of the volume and thus have reduced effects. In this work, we applied a linear interpolation in the conversion process from SR-PBI-HCT projections to virtual SR-PBI-CT projections to lessen the effect of defects (i.e. the cause of artifacts). As a result, SR-PBI-HCT has significantly reduced artifacts compared with SR-PBI-CT.

Theoretically, the use of a large-area detector and/or increased pitch would help to further reduce the helical artifacts because these can spread artifacts over a larger region of the volume. However, such settings may lead to other issues, for example, additional artifacts/noise problems if using a large-area detector (see Fig. 5[link] and related discussion), and streaking artifacts if using a pitch of more than 2 (see Fig. 8 and related discussion). As such, the imaging settings become complicated, particularly for the case of imaging live animals where the radiation dose is critical; as such, a trade-off has to be made between the radiation dose and imaging quality.

[Figure 5]
Figure 5
X-ray flux distribution comparison between SR-PBI-HCT and SR-PBI-CT. Flat images of SR-PBI-HCT (a1) and SR-PBI-CT (b1). Gray value profiles of the flat images (shown by the red lines) of SR-PBI-HCT (a2) and SR-PBI-CT (b2).

In addition, we noticed that the results from SR-PBI-HCT have higher values of SNR and CNR than those of SR-PBI-CT (even for areas without ring artifacts). One of the reasons behind this could be the different intensity/flux distribution ranges in the vertical direction due to different active detector height (i.e. 2.6 mm for SR-PBI-HCT and 9.31 mm for SR-PBI-CT). Fig. 5[link] shows the flat-field images and their vertical gray value profiles (i.e. flux distribution) obtained using SR-PBI-HCT and SR-PBI-CT.

From Figs. 5[link](a1) and 5[link](b1), SR-PBI-HCT only uses the central X-ray beam so that the overall intensity is stronger and the distribution is more uniform compared with the beam received by a larger detector in SR-PBI-CT. Specifically, for SR-PBI-HCT, the vertical X-ray flux intensity drop [\{{\rm drop} = [{\rm max}(I) - {\rm min}(I)] / {\rm max}(I)\}] is 11.43% (from center to sides), which is much smaller than the drop of 53.77% for SR-PBI-CT. Although the scanned scaffold has a smaller height (5 mm) than the active height of the detector (9.31 mm), the intensity differences still negatively affect the image quality (i.e. increased noise level in the relatively low-flux areas). This effect will become more severe with low-dose imaging for in vivo imaging. Generally, the smaller detector in SR-PBI-HCT overcomes the limitation of beam height for some cases and is also beneficial to control radiation dose distribution for in vivo imaging. Additionally, the small detector in SR-PBI-HCT reduces the possibility of including damaged pixels, thereby decreasing the possibility of introducing additional unexpected artifacts. Although objects can also be imaged with SR-PBI-CT using a small detector via multiple view scanning (by moving the sample longitudinally for the next scan), the intensity distributions are less uniform because overlapping only occurs at the margins between different views. The overlapping in SR-PBI-CT also increases the radiation dose compared with continuous scan mode in SR-PBI-HCT.

Fig. 6[link] displays the reconstructed results of the 3D QRM bar pattern phantom from SR-PBI-HCT and SR-PBI-CT for the analysis of spatial resolution. None of the PhR and ring artifact removal algorithms were applied to avoid introducing impacts on spatial resolution. From Figs. 6[link](a2) and 6[link](b2), the line pairs at the center region of the phantom show comparable spatial resolution reflected by the similarly discernible line pairs between SR-PBI-HCT and SR-PBI-CT. However, due to the effect of ring artifacts, we notice that there are obvious structural distortions on the line pairs for SR-PBI-CT, as indicated by the red arrows. The curves in Figs. 6[link](a3) and 6[link](b3) show the relationship between the MTF amplitude and spatial resolution [measured as line-pairs per millimetre (Lps mm−1)], as well as the corresponding cutoff resolution (10% MTF, i.e. smallest resolvable object) of 17.2 Lps mm−1 (i.e. linewidth: 29.07 µm) for SR-PBI-CT and 17.6 Lps mm−1 (i.e. linewidth: 28.41 µm) for SR-PBI-HCT. SR-PBI-HCT shows a slightly higher spatial resolution than SR-PBI-CT and the main reason may be the disruption of ring artifacts on SR-PBI-CT images. By combining the results from Figs. 4[link] and 6[link], SR-PBI-HCT shows promise of addressing the ring artifacts, while having no loss in spatial resolution.

[Figure 6]
Figure 6
Reconstructed 3D QRM bar pattern phantom using SR-PBI-CT and SR-PBI-HCT (gray scale: 0–255); (a1) SR-PBI-CT image and (b1) SR-PBI-HCT image and the corresponding (a2) SR-PBI-CT and (b2) SR-PBI-HCT ROIs enlarged from the yellow squares in (a1) and (b1); (a3)–(b3) corresponding MTF amplitudes with cutoff resolutions shown.

3.2. Influence of helical pitch of SR-PBI-HCT

The effect of p on hydro­gel scaffold images is shown in Fig. 7[link]. Overall, there is little visible effect on reconstructed 3D results of p [Figs. 7[link](a1)–7(d1)] with the range 1.3–1.9, and all the strands within the assessed ROIs are visibly recognizable [Figs. 7[link](a2)–7(d2)]. The gray value profiles of strands for all cases are very close to each other [Figs. 7[link](a3)–7(d3)]. Nevertheless, one can still observe the slight artifacts on images obtained with p = 1.3, as indicated by the red arrow [Fig. 7[link](a2)]. These artifacts are diminished for images with p = 1.5, 1.7 and 1.9. This is because higher p actually disperses the ring artifacts in a larger vertical region. As a result, the artifacts assigned to each SR-PBI-HCT slice will be reduced. In addition, SNR and CNR are improved when p increases from 1.3 to 1.5, but are then reduced when p increases from 1.5 to 1.9. The analysis of 3D spatial resolution [Figs. 7[link](a3)–7(d3)] with FSC also produces similar results. The Fourier image resolution (FIRE) is the reciprocal of spatial frequency at the intersection of the smoothed FSC curve with a correlation threshold of 1/7 (i.e. 0.14). One possible reason could be, for a reconstructed slice, the illumination range of X-rays (i.e. the total angle covered by the flux when samples are rotating) decreases with increasing p. A small illumination range usually produces worse image quality (Hayes et al., 2021[Hayes, J. W., Montoya, J., Budde, A., Zhang, C., Li, Y., Li, K., Hsieh, J. & Chen, G.-H. (2021). IEEE Trans. Med. Imaging, 40, 3077-3088.]). In summary, p = 1.5 can produce higher image quality because it balances the effects from ring artifact dispersion and illumination range. For a parallel beam geometry system, it should be realized that, when p > 2.0, overlapping between adjacent rotations will be incomplete, and this will lead to streaking artifacts, as shown in Fig. 8[link] (p = 2.5).

[Figure 7]
Figure 7
Effect of helical pitches on hydro­gel scaffold images using SR-PBI-HCT with pitches of 1.3, 1.5, 1.7 and 1.9; (a1)–(d1) reconstructed 3D results; (a2)–(d2) corresponding 2D images; (a3)–(d3) corresponding gray value profiles measured at the red line positions in (a2)–(d2); (a4)–(d4) 3D spatial resolution analyses with FSC, smoothed FSC and FIRE.
[Figure 8]
Figure 8
Hydro­gel scaffold images with a pitch of 2.5. (a1) Reconstructed slice; (a2) ROI [yellow square in (a1)]; (a3) example of converted virtual projections (gray regions are missing information between adjacent rotations).

3.3. Influence of photon energy on SR-PBI-HCT

The effect of E on hydro­gel scaffold images using SR-PBI-HCT is shown in Fig. 9[link]. The main differences associated with X-ray energies are the variations in image contrast, which generally involve the combination of phase contrast and absorption contrast in synchrotron radiation imaging. In the typical X-ray photon energy range 10–140 keV, the absorption coefficient scales as 1/E3 (Als-Nielsen & McMorrow, 2011[Als-Nielsen, J. & McMorrow, D. (2011). Elements of Modern X-ray Physics, 2nd ed. Chichester: Wiley.]) before or after the absorption edge. A lower E has a higher absorption coefficient and therefore is less penetrating. When E = 40 keV and 50 keV, the X-ray photons penetrate with less absorption, resulting in little difference in image contrast (Fig. 9[link]). Thus, the strands are hard to identify for E = 40 keV and 50 keV [Figs. 9[link](b2) and 9[link](c2)] and the edges of strands are not clear [Figs. 9[link](b1) and 9[link](c1), as indicated by the red arrows]. Besides, 3D spatial resolution analysis with FSC shows that both 40 keV and 50 keV have much larger values of FIRE than 30 keV [Figs. 9[link](b1) and 9[link](c1)]. Consequently, to obtain scaffold images with higher SNR and CNR using SR-PBI-HCT, 30 keV is optimal compared with 40 keV and 50 keV. Theoretically, with E < 30 keV, imaged objects have a higher absorption coefficient and therefore result in higher contrast; however, for our scaffolds, imaging in a tube with 50 ml water (in vitro, simulating in vivo imaging conditions for scaffolds implanted in a rat leg for nerve tissue regeneration), E < 30 keV results in higher noise level and higher radiation dose because a larger proportion of photons are absorbed.

[Figure 9]
Figure 9
Effect of photon energies on hydro­gel scaffold images using SR-PBI-HCT with 30 keV, 40 keV and 50 keV; (a1)–(c1) reconstructed 3D results; (a2)–(c2) corresponding ROI images; (a3)–(c3) corresponding gray value profiles measured at the red line positions in (a2)–(c2); (a4)–(c4) 3D spatial resolution analysis with FSC, smoothed FSC and FIRE.

3.4. Influence of projection numbers of SR-PBI-HCT

The effect of Np on hydro­gel scaffold images is shown in Fig. 10[link]. For Np = 500, 1000 and 3000, hydro­gel scaffolds present sufficient image contrast for segmentation and 3D volume rendering display [Figs. 10[link](a1)–10[link](c1)]. Although the noise is increased for Np = 500 and 1000 compared with Np = 3000, strands can still be clearly identified [Figs. 10[link](a2)–10[link](c2)], which is also supported by quantitative analysis with the value of FIRE [Figs. 10[link](a4)–10[link](c4)]. The radiation dose for Np = 500 is lower than for the two other cases (see Table 1). Therefore, Np = 500 is the optimum when considering image quality and radiation dose.

[Figure 10]
Figure 10
Effect of projection numbers on hydro­gel scaffold images using SR-PBI-HCT with Np = 500, 1000 and 3000; (a1)–(c1) reconstructed 3D results; (a2)–(c2) corresponding ROI images; (a3)–(c3) corresponding gray value profiles measured at the red line positions in (a2)–(c2); (a4)–(c4) 3D spatial resolution analysis with FSC, smoothed FSC and FIRE.

3.5. Low-dose imaging

Although the image quality is satisfactory with E = 30 keV and Np = 500, the 17.67 Gy radiation dose is high for in vivo live-animal imaging. We tested the feasibility of SR-PBI-HCT for hydro­gel scaffold imaging within an acceptable dose range [∼500 mGy for rat protocols (Pratt et al., 2014[Pratt, I., Belev, G., Zhu, N., Chapman, L. & Cooper, D. (2014). Phys. Med. Biol. 60, 211-232.])]. Specifically, a scaffold imaging experiment with 2 × 2 binning mode and Np = 250 was also conducted. In addition, an NDF with a thickness of 80 mm was used to reduce the photon flux. The measured radiation dose was 54.72 mGy s−1. Results (Fig. 11[link]) show that scaffolds can still be identified and segmented from the background (i.e. water) while the radiation dose (∼342 mGy) remains suitable for in vivo animal imaging. This study reveals that SR-PBI-HCT imaging, though lacking in vivo results currently, is promising for in vivo TERM applications.

[Figure 11]
Figure 11
Reconstruction results of a hydro­gel scaffold using SR-PBI-HCT with Np = 250 (2 × 2 binning): (a1) reconstructed 3D result; (a2) corresponding ROI image; (a3) corresponding gray value profile measured at the red line positions in (a2); (a4) 3D spatial resolution analysis with FSC, smoothed FSC and FIRE.

4. Conclusions

In TERM, visualization of low-density tissue scaffolds following their implantation is crucial, yet challenging. In this paper, we present a study on the integration of helical acquisition mode with SR-PBI-CT to non-invasively and non-destructively visualize/characterize low-density scaffolds. The results demonstrate the improvement of contrast and the significant advantage of avoiding ring artifacts without introducing additional artifacts. We also showed that p = 1.5, E = 30 keV and Np = 500 were suited for hydro­gel scaffold imaging in vitro using the current SR-PBI-HCT configuration of BMIT. In addition, with 2 × 2 binning mode and Np = 250, the SR-PBI-HCT can produce satisfactory results while the radiation dose (∼342 mGy, voxel size of 26 µm) remains suitable for in vivo animal imaging. The results obtained in this study reveal that the SR-PBI-HCT imaging method is a powerful tool for visualizing and characterizing hydro­gel scaffolds in terms of image quality and radiation dose, forming a solid base for in vivo 3D non-invasive characterization in TERM.

Supporting information


Acknowledgements

We acknowledge Dr Sergey Gasilov for his contribution to the development of the SR-PBI-HCT technique.

Funding information

The following funding is acknowledged: Natural Sciences and Engineering Research Council of Canada (grant nos. RGPIN 06007-2019; RGPIN 06396-2019). Experimental research described in this paper was performed at the Canadian Light Source, which is a national research facility supported by the Canada Foundation for Innovation (CFI), NSERC, National Research Council (NRC), Canadian Institutes of Health Research (CIHR), Government of Saskatchewan and the University of Saskatchewan.

References

First citationAls-Nielsen, J. & McMorrow, D. (2011). Elements of Modern X-ray Physics, 2nd ed. Chichester: Wiley.  Google Scholar
First citationBagherzadeh, R., Latifi, M., Najar, S. S., Tehran, M. A. & Kong, L. (2013). J. Biomed. Mater. Res. 101A, 765–774.  CrossRef CAS Google Scholar
First citationBartoš, M., Suchý, T. & Foltán, R. (2018). Biomed Eng. OnLine, 17, 110.  PubMed Google Scholar
First citationBawolin, N. K. & Chen, X. (2016). J. Funct. Biomater. 7, 17.  CrossRef PubMed Google Scholar
First citationCastelli, E., Tonutti, M., Arfelli, F., Longo, R., Quaia, E., Rigon, L., Sanabor, D., Zanconati, F., Dreossi, D., Abrami, A., Quai, E., Bregant, P., Casarin, K., Chenda, V., Menk, R. H., Rokvic, T., Vascotto, A., Tromba, G. & Cova, M. A. (2011). Radiology, 259, 684–694.  Web of Science CrossRef PubMed Google Scholar
First citationChapman, D., Thomlinson, W., Johnston, R., Washburn, D., Pisano, E., Gmür, N., Zhong, Z., Menk, R., Arfelli, F. & Sayers, D. (1997). Phys. Med. Biol. 42, 2015–2025.  CrossRef CAS PubMed Web of Science Google Scholar
First citationChen, C.-W., Betz, M. W., Fisher, J. P., Paek, A. & Chen, Y. (2011). Tissue Eng. C, 17, 101–112.  CrossRef Google Scholar
First citationChen, D. X. (2019). Extrusion Bioprinting of Scaffolds for Tissue Engineering Applications. Cham: Springer.  Google Scholar
First citationChen, J., Cai, Z., Wei, Q., Wang, D., Wu, J., Tan, Y., Lu, J. & Ai, H. (2020). J. Mater. Chem. B, 8, 316–331.  CrossRef CAS PubMed Google Scholar
First citationDuan, X., Li, N., Chen, X. & Zhu, N. (2021). Tissue Eng. C, 27, 573–588.  CrossRef CAS Google Scholar
First citationEdlow, B. L., Mareyam, A., Horn, A., Polimeni, J. R., Witzel, T., Tisdall, M. D., Augustinack, J. C., Stockmann, J. P., Diamond, B. R., Stevens, A., Tirrell, L. S., Folkerth, R. D., Wald, L. L., Fischl, B. & van der Kouwe, A. (2019). Sci. Data, 6, 244.  Web of Science CrossRef PubMed Google Scholar
First citationFedon, C., Rigon, L., Arfelli, F., Dreossi, D., Quai, E., Tonutti, M., Tromba, G., Cova, M. A. & Longo, R. (2018). J. Med. Imaging, 5, 013503.  Web of Science CrossRef Google Scholar
First citationFu, J., Willner, M., Chen, L., Tan, R., Achterhold, K., Bech, M., Herzen, J., Kunka, D., Mohr, J. & Pfeiffer, F. (2014). Phys. Med. 30, 374–379.  CrossRef PubMed Google Scholar
First citationFujita, H., Tsai, D.-Y., Itoh, T., Doi, K., Morishita, J., Ueda, K. & Ohtsuka, A. (1992). IEEE Trans. Med. Imaging, 11, 34–39.  CrossRef PubMed CAS Web of Science Google Scholar
First citationHagen, C., Diemoz, P., Endrizzi, M., Rigon, L., Dreossi, D., Arfelli, F., Lopez, F., Longo, R. & Olivo, A. (2014). Opt. Express, 22, 7989–8000.  CrossRef CAS PubMed Google Scholar
First citationHarrison, K. D., Sales, E., Hiebert, B. D., Panahifar, A., Zhu, N., Arnason, T., Swekla, K. J., Pivonka, P., Chapman, L. D. & Cooper, D. M. (2022). J. Bone Miner. Res. 37, 2244–2258.  CrossRef CAS PubMed Google Scholar
First citationHayes, J. W., Montoya, J., Budde, A., Zhang, C., Li, Y., Li, K., Hsieh, J. & Chen, G.-H. (2021). IEEE Trans. Med. Imaging, 40, 3077–3088.  CrossRef PubMed Google Scholar
First citationHeel, M. van & Schatz, M. (2005). J. Struct. Biol. 151, 250–262.  Web of Science PubMed Google Scholar
First citationIzadifar, M., Babyn, P., Chapman, D., Kelly, M. E. & Chen, X. (2017). J. Synchrotron Rad. 24, 842–853.  CrossRef IUCr Journals Google Scholar
First citationIzadifar, M., Babyn, P., Kelly, M. E., Chapman, D. & Chen, X. (2017). Tissue Eng. C, 23, 548–564.  CrossRef CAS Google Scholar
First citationIzadifar, Z., Honaramooz, A., Wiebe, S., Belev, G., Chen, X. & Chapman, D. (2016). Biomaterials, 82, 151–167.  CrossRef CAS PubMed Google Scholar
First citationKotecha, M., Magin, R. L. & Mao, J. J. (2017). Magnetic Resonance Imaging in Tissue Engineering. Hoboken: Wiley.  Google Scholar
First citationLongo, R., Tonutti, M., Rigon, L., Arfelli, F., Dreossi, D., Quai, E., Zanconati, F., Castelli, E., Tromba, G. & Cova, M. A. (2014). Philos. Trans. R. Soc. A, 372, 20130025.  Web of Science CrossRef Google Scholar
First citationMiao, H., Panna, A., Gomella, A. A., Bennett, E. E., Znati, S., Chen, L. & Wen, H. (2016). Nat. Phys. 12, 830–834.  CrossRef CAS PubMed Google Scholar
First citationMomose, A., Kawamoto, S., Koyama, I., Hamaishi, Y., Takai, K. & Suzuki, Y. (2003). Jpn. J. Appl. Phys. 42, L866–L868.  Web of Science CrossRef CAS Google Scholar
First citationMueller, K. M. A., Topping, G., Schwaminger, S. P., Zou, Y., Rojas-González, D. M., De-Juan-Pardo, E. M., Berensmeier, S., Schilling, F. & Mela, P. (2021). Biomater. Sci. 9, 4607–4612.  CrossRef CAS PubMed Google Scholar
First citationNaghieh, S. & Chen, X. (2021). J. Pharm. Anal. 11, 564–579.  CrossRef PubMed Google Scholar
First citationNesterets, Y. I., Gureyev, T. E. & Dimmock, M. R. (2018). J. Phys. D Appl. Phys. 51, 115402.  CrossRef Google Scholar
First citationNezhad-Mokhtari, P., Ghorbani, M., Roshangar, L. & Soleimani Rad, J. (2019). Eur. Polym. J. 117, 64–76.  CAS Google Scholar
First citationNing, L., Sun, H., Lelong, T., Guilloteau, R., Zhu, N., Schreyer, D. J. & Chen, X. (2018). Biofabrication, 10, 035014.  CrossRef PubMed Google Scholar
First citationNing, L., Zhu, N., Smith, A., Rajaram, A., Hou, H., Srinivasan, S., Mohabatpour, F., He, L., Mclnnes, A., Serpooshan, V., Papagerakis, P. & Chen, X. (2021). Appl. Mater. Interfaces, 13, 25611–25623.  CrossRef CAS Google Scholar
First citationOliva, P., Di Trapani, V., Arfelli, F., Brombal, L., Donato, S., Golosio, B., Longo, R., Mettivier, G., Rigon, L., Taibi, A., Tromba, G., Zanconati, F. & Delogu, P. (2020). Sci. Rep. 10, 17430.  CrossRef PubMed Google Scholar
First citationOlubamiji, A. D., Izadifar, Z. & Chen, D. X. (2014). Tissue Eng. B, 20, 503–522.  Web of Science CrossRef Google Scholar
First citationOlubamiji, A. D., Izadifar, Z., Zhu, N., Chang, T., Chen, X. & Eames, B. F. (2016). J. Synchrotron Rad. 23, 802–812.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationOlubamiji, A. D., Zhu, N., Chang, T., Nwankwo, C. K., Izadifar, Z., Honaramooz, A., Chen, X. & Eames, B. F. (2017). Tissue Eng. C, 23, 156–168.  CrossRef Google Scholar
First citationPaganin, D., Mayo, S. C., Gureyev, T. E., Miller, P. R. & Wilkins, S. W. (2002). J. Microsc. 206, 33–40.  Web of Science CrossRef PubMed CAS Google Scholar
First citationParrilli, A., Pagani, S., Maltarello, M., Santi, S., Salerno, A., Netti, P., Giardino, R., Rimondini, L. & Fini, M. (2014). J. Microsc. 255, 20–29.  CrossRef CAS PubMed Google Scholar
First citationPelt, D. M. & Parkinson, D. Y. (2018). Meas. Sci. Technol. 29, 034002.  Web of Science CrossRef Google Scholar
First citationPhipps, M. C., Clem, W. C., Grunda, J. M., Clines, G. A. & Bellis, S. L. (2012). Biomaterials, 33, 524–534.  CrossRef CAS PubMed Google Scholar
First citationPratt, I., Belev, G., Zhu, N., Chapman, L. & Cooper, D. (2014). Phys. Med. Biol. 60, 211–232.  CrossRef PubMed Google Scholar
First citationRaven, C. (1998). Rev. Sci. Instrum. 69, 2978–2980.  Web of Science CrossRef CAS Google Scholar
First citationStachewicz, U., Szewczyk, P. K., Kruk, A., Barber, A. H. & Czyrska-Filemonowicz, A. (2019). Mater. Sci. Eng. C, 95, 397–408.  CrossRef CAS Google Scholar
First citationSuzuki, Y., Yagi, N. & Uesugi, K. (2002). J. Synchrotron Rad. 9, 160–165.  Web of Science CrossRef IUCr Journals Google Scholar
First citationTaba, S. T., Gureyev, T. E., Alakhras, M., Lewis, S., Lockie, D. & Brennan, P. C. (2018). Am. J. Roentgenol. 211, 133–145.  Web of Science CrossRef Google Scholar
First citationTavakoli Taba, S., Baran, P., Lewis, S., Heard, R., Pacile, S., Nesterets, Y. I., Mayo, S. C., Dullin, C., Dreossi, D., Arfelli, F., Thompson, D., McCormack, M., Alakhras, M., Brun, F., Pinamonti, M., Nickson, C., Hall, C., Zanconati, F., Lockie, D., Quiney, H. M., Tromba, G., Gureyev, T. E. & Brennan, P. C. (2019). Acad. Radiol. 26, e79–e89.  CrossRef PubMed Google Scholar
First citationTeodori, L., Crupi, A., Costa, A., Diaspro, A., Melzer, S. & Tarnok, A. (2017). J. Biophoton, 10, 24–45.  CrossRef Google Scholar
First citationThompson, A. C. & Vaughan, D. (2001). X-ray Data Booklet. Lawrence Berkeley National Laboratory, University of California Berkeley, CA, USA.  Google Scholar
First citationVitas, S., Segmehl, J. S., Burgert, I. & Cabane, E. (2019). Materials, 12, 416.  CrossRef PubMed Google Scholar
First citationVo, N. T., Atwood, R. C. & Drakopoulos, M. (2018). Opt. Express, 26, 28396–28412.  Web of Science CrossRef PubMed Google Scholar
First citationVogelgesang, M., Farago, T., Morgeneyer, T. F., Helfen, L., dos Santos Rolo, T., Myagotin, A. & Baumbach, T. (2016). J. Synchrotron Rad. 23, 1254–1263.  Web of Science CrossRef IUCr Journals Google Scholar
First citationWang, L., Xu, M., Luo, L., Zhou, Y. & Si, P. (2018). Sci. Rep. 8, 2802.  CrossRef PubMed Google Scholar
First citationWernick, M. N., Wirjadi, O., Chapman, D., Zhong, Z., Galatsanos, N. P., Yang, Y., Brankov, J. G., Oltulu, O., Anastasio, M. A. & Muehleman, C. (2003). Phys. Med. Biol. 48, 3875–3895.  Web of Science CrossRef PubMed Google Scholar
First citationWysokinski, T. W., Chapman, D., Adams, G., Renier, M., Suortti, P. & Thomlinson, W. (2007). Nucl. Instrum. Methods Phys. Res. A, 582, 73–76.  Web of Science CrossRef CAS Google Scholar
First citationYao, L., Zeng, D., Chen, G., Liao, Y., Li, S., Zhang, Y., Wang, Y., Tao, X., Niu, S., Lv, Q., Bian, Z., Ma, J. & Huang, J. (2019). Phys. Med. Biol. 64, 035018.  CrossRef PubMed Google Scholar
First citationYou, F., Wu, X., Zhu, N., Lei, M., Eames, B. F. & Chen, X. (2016). ACS Biomater. Sci. Eng. 2, 1200–1210.  CrossRef CAS PubMed Google Scholar
First citationZhu, N., Chapman, D., Cooper, D., Schreyer, D. J. & Chen, X. (2011). Tissue Eng. C, 17, 1071–1080.  Web of Science CrossRef CAS Google Scholar
First citationZhu, X., Hu, Y., Wu, G., Chen, W. & Bao, N. (2021). ACS Nano, 15, 9273–9298.  CrossRef CAS PubMed Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775
Follow J. Synchrotron Rad.
Sign up for e-alerts
Follow J. Synchrotron Rad. on Twitter
Follow us on facebook
Sign up for RSS feeds