Coaxial electrospun biomimetic copolymer fibres for application in diffusion magnetic resonance imaging

Objective. The use of diffusion magnetic resonance imaging (dMRI) opens the door to characterizing brain microstructure because water diffusion is anisotropic in axonal fibres in brain white matter and is sensitive to tissue microstructural changes. As dMRI becomes more sophisticated and microstructurally informative, it has become increasingly important to use a reference object (usually called an imaging phantom) for validation of dMRI. This study aims to develop axon-mimicking physical phantoms from biocopolymers and assess their feasibility for validating dMRI measurements. Approach. We employed a simple and one-step method—coaxial electrospinning—to prepare axon-mimicking hollow microfibres from polycaprolactone-b-polyethylene glycol (PCL-b-PEG) and poly(D, L-lactide-co-glycolic) acid (PLGA), and used them as building elements to create axon-mimicking phantoms. Electrospinning was firstly conducted using two types of PCL-b-PEG and two types of PLGA with different molecular weights in various solvents, with different polymer concentrations, for determining their spinnability. Polymer/solvent concentration combinations with good fibre spinnability were used as the shell material in the following co-electrospinning process in which the polyethylene oxide polymer was used as the core material. Following the microstructural characterization of both electrospun and co-electrospun fibres using optical and electron microscopy, two prototype phantoms were constructed from co-electrospun anisotropic hollow microfibres after inserting them into water-filled test tubes. Main results. Hollow microfibres that mimic the axon microstructure were successfully prepared from the appropriate core and shell material combinations. dMRI measurements of two phantoms on a 7 tesla (T) pre-clinical scanner revealed that diffusivity and anisotropy measurements are in the range of brain white matter. Significance. This feasibility study showed that co-electrospun PCL-b-PEG and PLGA microfibre-based axon-mimicking phantoms could be used in the validation of dMRI methods which seek to characterize white matter microstructure.


Introduction
Soft tissues such as human brain white matter and cardiac muscle exhibit a preferred microstructural orientation, with axonal fibres of 0.16-9 μm [1] and cardiac myocytes of 10-25 μm [2] in diameter. Diffusion magnetic resonance imaging (dMRI) has become a non-invasive tool to characterise such microstructure, due to the fact that water diffusion is anisotropic in these fibrous tissues and is sensitive to tissue microstructural changes [3]. Physical phantoms have been widely used for validation of dMRI 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t methods applied to neurology, such as tractography and microstructure measurement [4]. Melt-spun hollow microfibres [5,6] and 3D printed fibres [7] have shown potential for use in the construction of axon-mimicking MR phantoms. Despite the promise of these approaches, they suffer from a number of disadvantages. Melt-spun hollow polypropylene (PP) fibres have ~12 µm inner diameter and are mono-dispersed [5], whereas fused deposition modelling (FDM) 3D printed fibres have a thickness of 0.2 mm [7], neither of which provides a realistic mimic of axonal dimensions. The construction processes of both melt-spun microfibre-based phantoms and 3D printed phantoms involves multiple steps and require high temperature during processing. For instance, preparing PP fibre phantom involves melt spinning at 210 degrees Celsius(C) and then the manual assembly of tens of thousands of filaments into small chambers ~10 square millimetres in size [8]; the fibre packing density and orientation in the hollow PP fibre phantoms are manually controlled after melt-spinning [6]; 3D printed phantom construction involves 3D printing of an ink composed of polyvinyl alcohol (PVA)+rubber elastomer at high temperature of 230 degrees C and PVA removal via sonic water rinsing [9]; the 3D printing process allows automation of the process of generating MRI phantoms with controllable fibre orientation [9]. In addition, the polymer used to generate melt-spun axon-mimicking fibres, namely polypropylene (PP), is highly hydrophobic, which has made water-filling process labour-intensive in PP-derived fibre phantoms [6]. Most recently, hydrophilic Nylon hollow microfibres with 0.8 to 5 µm in diameter were prepared via bi-component melt-spinning for the construction of axon-mimicking phantoms [10,11]. These hollow microfibres allow water to enter the phantoms more easily than those PP microfibres but involve using highly sophisticated spinneret and a two-step procedure. Two high temperature involved melt spinning and FDM 3D printing techniques are not "green" processes.
Co-electrospun hollow polycaprolactone (PCL) fibres have been produced in the approximate range 1 to 13 µm [12,13], providing a fibre diameter distribution which mimics in vivo microstructure more closely than other phantom materials developed to date. The formation of axon-mimicking hollow microfibres via co-electrospinning involves fabrication in a one-step process at room temperature [12,14]. Fibre packing and orientation is mainly determined by the jet deposition process, which can be tuned in a process often called direct-jet co-electrospinning [15]. This can be achieved by optimizing the translation speed of an x-y translating mechanical collector [16] or introducing a pair of auxiliary electrodes to the x-y collector [17]. In addition, co-electrospun hollow hydrophobic fibre phantoms (e.g. PCL) can readily be made hydrophilic by adding a surfactant in the spinning solution in the coelectrospinning process [18]. Overall, co-electrospinning has several advantages over melt-spinning and 3D printing technologies in terms of available polymers, relative simplicity of one-step formation at room temperature and tuning of microstructural and chemical properties.
Polycaprolactone (PCL) and poly(D,L-lactide-co-glycolic) acid (PLGA, Fig. 1a) are FDA-approved biopolymers that are widely processed via electrospinning or electrospraying into fibres or particles for biomedical applications such as tissue engineering and drug delivery [19,20]. PCL is semicrystalline and hydrophobic polymer; PLGA is a copolymer of poly lactic acid (PLA) and poly glycolic acid (PGA). Glycolic acid is more hydrophilic than lactic acid, and therefore a higher ratio of glycolic acid-to-lactic acid increases the hydrophilicity of the polymer. Poly(ethylene glycol) (PEG) is a water soluble polymer and has also received approval from the FDA for application mainly in drug delivery [21]. PEG is often combined with several polymers, such as PCL-b-PEG (Fig. 1b), polylactide (PLA), poly(glycolic acid) (PGA) and their co-polyesters (PLGA) to increase their biocompatibility and to allow them to be used as drug carriers. For instance, to improve the hydrophilicity and regulate the biodegradation rate of PCL, random and block copolymers of PCL with PEG can be prepared. The PCL-b-PEG copolymer has been widely applied in the design of drug delivery systems [22][23][24]. Due to their ease of processing by electrospinning and tuneable hydrophilicity, these two co-polymers may have great potential in creating water-filled axon-mimicking phantoms for the validation of diffusion MRI .  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t Here we report for the first time the one-step formation of hollow microfibres from PCL-b-PEG and PLGA via co-electrospinning, as well as the feasibility of using these fibres to create axon-mimicking MRI phantoms. Electrospinning is firstly conducted using PCL-b-PEG and PLGA polymers. The solution properties (e.g. polymer molecular weight, solvent and solution concentration) and process parameters (e.g. applied voltage, working distance and flow rate) are optimized to produce smooth fibres from these two polymers. The polymer solutions that show good fibre spinnability are used as the shell material in the subsequent co-electrospinning process in which the highly spinnable polyethylene oxide (PEO) polymer is used as the core material. Optical and electron microscopy is used to determine the microstructure of both electrospun and co-electrospun fibres. Hollow microfibres that mimic brain axon microstructure are successfully prepared from the PEO core and PCL-b-PEG or PLGA shell material combinations, and are further collected in a uniaxially aligned form on a rotating collector. Two axon-mimicking MRI phantoms are constructed from these anisotropic hollow microfibres after inserting them into water-filled test tubes and are tested on a 7 tesla pre-clinical MR scanner, in order to determine water diffusion behaviour in these two phantoms.

Electrospinning of PCL-b-PEG and PLGA polymers
PCL-b-PEG400 was dissolved in a mixed solvent of CHCl 3 and DMF at concentrations of 20 wt.%, 25 wt.% and 30 wt.% (sets 1-3, Table 1) to determine the spinnability of the PCL-PEG copolymer. PCL-b-PEG1000 was also dissolved in a mixed solvent of CHCl 3 /DMF at concentrations of 10 wt.% and 15 wt.% (set 4, Table 1). PLGA-I was dissolved in CHCl 3 /DMF with variable blending ratios (sets 5-7, Table 1) and in 5/5 w/w THF/DMF (set 8, Table 1) at a concentration of 20 wt.% . To further explore the effect of the solvents used above, PLGA-II polymer, which has a higher molecular weight than PLGA-I, was dissolved in 8/2 w/w CHCl 3 /DMF at concentration of 20 wt.% (set 9, Table 1) and 5/5 w/w THF/DMF with three concentrations of 10, 15 to 20 wt.% (set 10, Table 1). These PCL-b-PEG and PLGA polymer solutions were processed under various process parameters listed in Table 1 into fibres on an electrospinning setup schematically shown by Fig.1(a) below and described previously [12]. The investigated polymer solutions and the corresponding process parameters are summarized in Table 1. In brief, a mixed solvent of CHCl 3 /DMF (8/2, w/w) was used a b A c c e p t e d M a n u s c r i p t to dissolve PCL-b-PEG at concentrations of 20, 25 and 30 wt.%. PLGA-I and -II were also dissolved in CHCl 3 /DMF (8/2, 7/3 and 5/5, w/w) and THF/DMF (5/5, w/w) at the concentrations of 10, 15, and 20 wt.%. A high-voltage power supply (PS/FC30R04.0e22, Glassman High Voltage, UK) was used to tune the applied voltage between 0 and 30 kV. A 10 mL plastic syringe with a stainless-steel needle (inner diameter 1.19 mm) mounted on a syringe pump (SP230IWZ, Multi-Syringe Pump, World Precision Instruments, UK) was used to feed polymer solution to the needle tip at a controllable feed rate (see Table 1 below). The fabricated fibres were then collected either on a glass slide on the top of a grounded metal plate for optical microscopy or on grounded aluminium foil for scanning electron microscopy (SEM). All experiments were conducted in a fume cupboard under ambient conditions. The corresponding cone-jet state observed by the naked eye for the first 10-15 min in each operatiaon and the resultant fibre morphogy based on optical or electron microcopy are also briefly describled in Table 1.    1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59

Co-electrospinning of PCL-b-PEG and PLGA polymers
The shell/core solutions and process parameters for co-electrospinning are given in Table 2.
Specifically, 15 wt.% PCL-b-PEG1000/CHCl 3 +DMF and 4 wt.% PEO/water were used as shell and core solutions in the co-electrospinning process. The co-electrospinning of PLGA was first conducted using 20 wt.% PLGA-I in CHCl 3 +DMF (8/2, w/w) as the shell fluid while the core solution was varied from 4 wt.% PEO in water to 2 wt.% PEO in CHCl 3 . In the case of a 4 wt.% PEO in water core, two settings for applied voltage/working distance (15.6 kV/13.5 cm and 10.5 kV/5 cm) were used to observe the jet behaviour while the core/shell flow rates were kept at 0.5/1.0 mL/h. The experimental set-up schematically shown in Fig. 2b used for preparing hollow microfibres has been described previously [12]. In brief, a coaxial spinneret, with two concentric needles (ID 0.41 and 1.19 mm), was filled with a shell solution of PCL-b-PEG or PLGA (outer needle) and a core solution of PEO (inner needle). The outer needle was connected to the positive electrode from a DC high voltage power supply (as was the case for the electrospinning above). The fibre collector, which was placed below the tip of the concentric needles, was connected to the grounded electrode.
Those parameters listed in Table 2, guaranteeing a reasonably stable co-electrospinning process, were reached after a series of optimizations with respect to the solvent system, polymer concentrations, flow rate and applied voltage. A wide rotating drum (~11 cm in diameter), which was spun at controllable revolutions per minute (rpm), was used as a collector to align the fibres. Once the inner core solution in the deposited fibres evaporated, which takes a few milliseconds [26], the hollow fibres were left with a solidified outer sheath. All experiments were conducted in a fume cupboard at ambient conditions. A solution of PCL-b-PEG or PLGA polymer in CHCl 3 /DMF or THF/DMF solvent was used as the shell fluid and PEO in deionized water or CHCl 3 acted as the core fluid. These two liquids were fed at a constant flow rate independently controlled by two syringe pumps. A c c e p t e d M a n u s c r i p t

Optical and scanning electron microscopy
The morphology of electrospun and co-electrospun fibres was observed using an Olympus BH2-UMA optical microscope and a Philips XL30 FEG scanning electron microscope (SEM) or a Phenom G2 pro desktop SEM with an accelerating voltage of 5 kV. In order to quickly determine the morphology of electrospun products, a glass slide was introduced into the electrospinning/co-electrospinning setup and held in contact with the grounded collector, in the centre of the spinning zone for a short period of time (usually less than one minute) at the beginning of the process. These were imaged using optical microscopy. The electrospun fibres collected on aluminium foil were coated with a thin layer of palladium-gold with a thickness of approximately 10 nm to increase their conductivity before SEM imaging. For imaging of fibre cross sections using SEM, fibres were first cut by sharp scissors in liquid nitrogen. For the quantitative analysis of the pores of co-electrospun hollow fibres, ImageJ was used for analysis based on a previously reported semi-automated method with Size and Circularity setting to be 0.5-15 µm and 0.001-1.00, respectively in "Analyze Particle" feature [16].

dMRI phantom construction
One PCL-b-PEG fibre phantom sample from the co-electrospun hollow PCL-b-PEG fibre strip (set 1, Table 2) and one PLGA fibre phantom sample were constructed from hollow PLGA fibre strip (set 5, Table 2), respectively. Two layers of PCL-b-PEG fibre strips of ~2.0 mm thick and 6 layers of PLGA fibre strips of ~0.35 mm thick, respectively, were packed into two 12 mm outer diameter (O.D.) glass tubes (Wilmad-LabGlass) that were filled with deionized water.

Results and discussion
The present work primarily reported the microstructures of copolymer fibres via electrospinning and co-electrospinning, and water diffusion behaviour in hollow microfibres, but nevertheless wettability and physico-chemical properties of copolymers as reported in previous studies [18,25] are important for the resultant fibre phantoms.

Electrospinning of PCL-b-PEG copolymers
Solution dripping was observed and became less frequent with increasing applied voltage or decreasing flow rate (set 1, Table 1). As shown in Fig. 3a-b, mixed structures comprising beaded PCL-b-PEG fibres with diameters up to a few microns and bead-free fibres with much thinner diameters (a few hundred nanometres) are present in the product from a 20 wt.% solution; the processes in sets 2-3 ( Table 1) were stable; but fibres from 25 wt% (Fig. 3c-e) and 30 wt.% (Fig. 3f-h) solutions are relatively smooth and nearly bead-free but micro-sized fibres are more likely to be formed from 30 wt.% concentration especially when a relatively high flow is used.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t Solution dripping occurred to 10 wt.% but not to 15 wt.% solution in set 4, Table 1; beaded fibres and bead-free fibre formed from 10 wt.% (Fig. 4a-g), and 15 wt.% (Fig. 4h-n), respectively. As shown in Fig. 4a-g, beaded PCL fibres are present in the products electrospun from the 10 wt.% solution regardless of the changing process parameters; fibres produced from 15 wt.% are bead-free ( Fig. 4hn) and have much larger sizes than those obtained from 10 wt.%. When the high flow rate of 1.5 mL/h was used, the resultant fibres (Fig. n) look much thicker (up to ca. 10 µm) than those produced under other conditions, and tend to merge. These thicker and merged fibres are expected to be caused by the insufficient jet stretching in the electrospinning process.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t PCL-b-PEG copolymers are synthesised by grafting active sites of hydroxyl groups to PCL chains via aminolysis, leading to products with controllable molecular weight and significantly improved water wettability of PCL alone [25]. The PCL-b-PEG400 and PCL-b-PEG1000 polymers are here for the first time demonstrated to be eletrospinnable under appropriate combinations of solution concentration and key process parameters. The combination of wettability and spinnability for PCL-b-PEG makes it a potentially useful material for creating axon-mimicking phantoms via co-electrospinning [18]. Due to the fibre size and lack of beading, 15 wt.% PCL-b-PEG1000 in CHCl 3 was chosen here to be used as the shell solution in the co-electrospinning of axon-mimicking fibres in Section 3.3.

Electrospinning of PLGA copolymers
An increase in applied voltage from 13.5 to 17 kV or a decrease in flow rate from 0.3 to 0.05 mL/h helped achieve a stable process. All the resultant PLGA-I fibres are bead-free and uniform in size (a few microns in diameter) (Supplementary information, Fig. S1a-d), both when the applied voltage is changed from 13.5 kV to 17.0 kV and the flow rate was reduced from 0.3 mL/h to 0.05 mL/h in set 5. This indicates that the spinnability of the 20 wt.% solution PLGA-I in CHCl 3 /DMF (8/2, w/w) is good, allowing fibre formation within a range of electric field strengths and/or flow rates. Solution was slightly dripping when changing the blend ratio of CHCl 3 /DMF from 8/2 to 7/3 w/w (set 6). There is an obvious change in morphology from smooth to beaded fibres on which the bead size could reach ca. 40 µm diameter (Supplementary information, Fig. S1e). Solution dripping disappeared when the proportion of DMF increased (CHCl 3 /DMF ratio from 7/3 to 5/5 w/w, set 7). Beaded fibres become dominant (Supplementary information, Fig. S1f ). A similar morphology transition caused by the solvent composition was observed in a previous study [27]. Both CHCl 3 and DMF are good solvents for the PLGA polymer but DMF has a higher evaporation rate and surface tension, which favours the formation of the beads or beaded fibres. When 5/5 w/w THF/DMF was used as the solvent (set 8), a stable process was achieved. PLGA fibres become defect-free and smoother (Supplementary information, Fig. S1g) than those fibres from 5/5 w/w CHCl 3 /DMF (Supplementary information, Fig.  S1f), indicating that the presence of THF favours the formation of smooth PLGA fibres compared with CHCl 3 when the blending ratio is kept at 5/5 w/w. This result is consistent with a previous study in which PLGA in THF/DMF formed smoother fibres than those from a PLGA solution in CHCl 3 /DMF  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t with the same polymer concentration [27]. This could be explained by the fact that the surface tension of PLGA in CHCl 3 /DMF is significantly higher than THF/DMF, which can promote the formation of beaded fibres from the former case [27].
As with the PCL-b-PEG copolymer, these results demonstrate that both PLGA-I and II can also be processed into smooth and uniform fibres using appropriate solvent systems. Diffusion MRI has revealed that water can penetrate into tumour cell-mimicking phantom comprising hollow PLGA microspheres of [28]. Therefore, the spinnability and wettability of PLGA polymers could enable them to be another suitable shell material in the co-electrospinning of axon-mimicking fibres.

Co-electrospinning of PCL-b-PEG and PLGA hollow microfibres
Previous study has demonstrated that the co-electrospinning process of 10 wt.% PCL/CHCl 3 +DMF as shell and 4 wt.% PEO/water as core is appropriate to produce hollow PCL microfibres with various sizes that can be used to create axon mimicking phantoms [12,18,29]. Similar experimental parameters (except with applied voltage of 15 kV) to those in these earlier studies were adopted in the co-electrospinning of a 4 wt.% PEO/water core and the 15 wt.% PCL-b-PEG1000/CHCl 3 +DMF shell to generate an aligned fibre bundle on a rotating drum. As shown in Fig.5a-c, the resultant PCL-b-PEG fibres in the bundle are predominantly aligned (as indicated by the arrow) and hollow.
In both processes the compound fluid jet was found to be stable and straight over an observation period of ~1 h. In the case of 2 wt.% PEO in CHCl 3 core, a straight and stable fluid jet was also achieved for an operation period of ~1 h 45 min. However, over a longer period of co-electrospinning, solidified PLGA polymer was seen gradually accumulating on the needle tip and finally blocked the needle, resulting in short interruptions of the process stability. The co-electrospinning process resumed and was stable after manually removing the solidified polymer from the needle tip. Fig. 5d-f, the resultant PLGA-I fibres generated from both PEO in water and PEO in CHCl 3 cores have a flat ribbon structure. A representative cross-sectional micrograph in Fig. 5g further reveals that PLGA-I fibres do not have hollow and circular structure. A change in the blending ratio of CHCl 3 /DMF from 8/2 to 7/3 or 5/5 w/w (giving a lower evaporation rate) was found to prevent the solidification of the PLGA solution. However, the decreasing CHCl 3 content favoured the formation of beaded fibres (Fig. 5h) collected on a glass slide in the co-electrospinning. The use of 5/5 THF/DMF solvent for PLGA-I was found to be very effective in changing the morphology from beaded to smooth fibres (Fig. 5i). This morphological change is not surprising because in electrospinning beaded fibres were generated from 20 wt.% PLGA-I in 7/3 or 5/5 w/w CHCl 3 /DMF (Fig. 4e and f) but smooth fibres were from 20 wt.% PLGA-I in 5/5 w/w THF/DMF (Fig. 4g). For the combination of a PEO in CHCl 3 core and PLGA-I in 8/2 w/w CHCl 3 /DMF shell, a change in the shell flow rate from 1 to 0.5 mL/h, while other governing process parameters were unchanged (except the collector speed), led to changes in the fibre morphology from collapsed (Fig. 5f) to hollow (Fig. 6a). When the shell was changed to be PLGA-I in 5/5 w/w THF/DMF and other parameters were constant, hollow and aligned PLGA fibres were also seen (Fig. 6b). For the shell solution of PLGA-II in 8/2 CHCl 3 /DMF or 5/5 THF/DMF Page 10 of 17 AUTHOR SUBMITTED MANUSCRIPT -BB-102476.R2 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t solvent, the PLGA fibres in the resultant strips were also hollow ( Fig. 6c and d) when the same core solution and process parameters were used, except the applied voltage used for PLGA-II in 5/5 THF/DMF. It should be mentioned however that the hollow PLGA fibres in these strips merge with their neighbours and there are also extra-fibre spaces of larger size than the interior of the hollow PLGA fibres, especially for the 8/2 CHCl 3 /DMF solvent. The cross-sections of PLGA fibre strips from THF/DMF (Fig. 6b and d) appear much denser than those from CHCl 3 /DMF (Fig. 6a and c).

MR imaging of water-filled PCL-b-PEG and PLGA fibre phantoms
The dMRI signal is sensitive to the diffusion of water molecules in the phantoms and reflects the extent to which the phantom microstructure restricts or hinders molecular motion. Fig. 8 shows the MD and FA maps of two fibre phantoms and one free water phantom in a 5 mm glass tube with corresponding colour bars. It can be clearly seen from the MD map (Fig. 8a) that the diffusion coefficient in the phantoms is lower than free water, indicating water diffusion in the phantoms is restricted and/or hindered. The FA map (Fig. 8b) provides evidence of anisotropy in both the PCL-b-PEG and PLGA fibre phantoms, which is expected given the uniaxially aligned fibre strips (Fig. 5a-c and Fig. 6c-d).
Page 13 of 17 AUTHOR SUBMITTED MANUSCRIPT -BB-102476.R2 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t  Table  2); Bottom right tube -PLGA fibre phantom (set 5, Table 2 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 A c c e p t e d M a n u s c r i p t Fig. 8c-e shows the MD and FA values extracted from ROIs in each phantom. It is clear that the free water phantom has narrower MD and FA distributions (Fig. 8c) than both hollow fibre phantoms due to its homogeneity. Free water has an MD value of 2.16 µm 2 /ms, in close agreement with literature values [37] and an FA value of 0.12, which is expected to be close to zero due to isotropic diffusion in the water-filled tube. The discrepancy in MD and non-zero FA between two free water phantoms could be attributed to the low number of diffusion directions, only 6. FA is known to be overestimated due to noise, making isotropic diffusion appear to have a low level of anisotropy [38], which explains the non-zero FA in the water-filled tube. The PCL-b-PEG fibre phantom has narrower distributions of MD and FA (Fig. 7e) than those of the PLGA fibre phantom (Fig. 7d), though the corresponding median values of MD and FA of these two fibre phantoms are very similar. As shown in the MD and FA maps ( Fig. 7a-b), in the PLGA fibre phantom sample there are six layers of strips in which gaps between fibre strips are present; in the PCL-b-PEG fibre strips there are only two fibre layers without obvious gaps present. The inhomogeneous structure in the PLGA fibre phantom can result in the partial volume effects where both fibres and free water contribute to the signal in voxels at the interface, and thus affect dMRI measurements. This is consistent with previous observations, where decreased values of FA were observed in partial voluming of free water [39] or free cyclohexane [40] filled fibre phantoms. The partial volume could contribute to the relatively wider distributions of MD and FA values of the PLGA fibre phantom. This effect could be minimized by tighter packing of the fibre strips, using thicker fibre strips, or increasing the image resolution. In addition, it should be noted that the values of MD and FA vary with diffusion time [37]. The diffusion time used here was much shorter than that typically used on a clinical scanner, which makes it difficult to compare the values of MD and FA with those from typical clinical MR scans. It is expected that with longer diffusion times, MD of these phantoms will decrease, and FA will increase. It can be envisaged that that MD and FA values of these copolymer fibre phantoms are sensitive to pore sizes and fibre orientation, as demonstrated in previous PCL fibre phantoms [29,41,42], but here the primary discussion only focused on two copolymers' feasibility as MRI phantom material. In addition, PCL fibre phantoms have high reproducibility [29,42]. We expect this reproducibility could also be achieved with the current co-electrospun phantom fibres. Future research will serve to reveal the relationship between MRI measurements with two copolymer fibres' microstructures.