【正文】 re size and porosity can be calculated by counting the average cellnumber and size of several SEMimages from one sample. One cut area with certain size was chosen and all pores were measuredmanually with the help of software of digital microscope (VHX500, Keyence Corporation, Osaka, Japan). The average diameter ofpores was calculated as Dmeasured. Due to the fact that the pores shownin the micrographs are 2D projections of 3D objects, their maximumdiameter may not be represented in the image. Following equation wasused for determination of the maximum spherical diameter, namedcorrected median pore diameter, from the measured pore diameter: [1]. （ 1） MicroCT (SkyScan 1172, SkyScan, Kontich, Belgium) was used toquantitatively measure the porous interconnectivity of implants – three8mm_11mm cylindrical samples from each implant (n 3) at 7 mmresolution using a voltage of 59 kV, and a current of 167 mA. Imagereconstruction and analysis were conducted using the software packageprovided by SkyScan. Samples were rotated 1808 around their long axisand three absorption images were recorded every of raw images of the samples were reconstructed with the standardSkyScan reconstruction software (NRecon) to serial coronalorientedtomograms using 3D cone beam reconstruction algorithm. For thereconstruction, beam hardening was set to 20% and ring artefactreduction to 12. The image analysis of the reconstructed axial bitmap images wasperformed using the standard SkyScan software (CTan and CTvol).First, a thresholding analysis was performed to determine the thresholdvalue for which the greyscale tomograms of scaffolds were mostaccurately represented by their binarised counterparts in terms ofporosity. The threshold value was set between 65 and 225 for this noise was removed by the ‘despeckling’ function. All objectssmaller than 500 voxels and not connected to the 3D body were thusremoved prior to further analysis. In order to eliminate potential edgeeffects, a cylindrical volume of interest (VOI) with a diameter of 5mmand a height of was selected in the centre of the scaffold. Scaffoldporosity was then calculated as follows: = 10 0% v ol .% of b i na r i se d ob j e c t ( sc a f f ol dm a t e r i a l ) i n V oiPorosity （ 2） All images underwent 3D analysis, followed by the quantification ofinterconnectivity using the ‘shrinkwrap’ function, which allows measuringthe fraction of pore volume in a scaffold that was accessible fromthe outside through openings of a certain minimum size [8]. A shrinkwrapprocess was performed between two 3D measurements to shrinkthe outside boundary of the VOI in a scaffold through openings the sizeof which was equal to or larger than a threshold value (0–280 mm wereused in this study). Interconnectivity was calculated as follows: DD ? （ 3） RESULTS AND DISCUSSION The SEM images (Figure 3) show the pore structures of foamedimplants from two molds in the injection speed variation with value of30 mm/s, when the other process parameters were kept unchanged(weight reduction of 35%, plasticizing temperature of 1808C, plasticizingpressure of 180 bar, mold temperature of 258C, and gas content of 2%). Itwas found that the left image, which came from the foamed implantfrom mold B, showed a significant larger pore size than right image frommold A. The interconnective pore size [9,10] which means the windowbetween two connective pores has also the same change trend. FromFigure 3 it could be qualitatively seen that the implants from mold B hada larger pore size and interconnective pore size and possibly had a higherporosity than those from mold A at the same process parameter. Figure 3. Different pore structures of mold B (left) and mold A (right) at the injectionspeed of 30 mm/s. Figure 4. Differences of the porosity at injection speed variation. s h r i n k w r a pmV= 1 0 0 %V VV? ?InterconnectivityIt was found from Figure 4 that the implants from mold B at everydifferent injection speed had a higher porosity than the implants frommold A. The porosity range of implants from mold B was between 73%and 79%, whereas by mold A this porosity range was between 60% and67%. At the same time the standard deviation of the porosity from moldB was significantly smaller than the deviation by mold A. Figure 5. The mean pore size from two molds at different injection speeds Figure 5 shows the mean pore size of two molds by different injectionspeeds. The pore size decreased with rise of the injection speed for twomolds. The same result was also found by other study [11]. The porediameter of the implants from mold B decreased from 340_17 mm to246_20 mm with injection speed increase。 the mold A showed the porediameter from 234_90 mm to 152_34 mm by the same injection speedvariation. The mean pore size from mold B at every speed was alsohigher pared with mold A. It was clear that the standard deviationfrom mold B was also significantly smaller than the values from mold A. Figure 6 shows the interconnective pore size of foamed implants. Theinterconnective pore size is very important for the tissue in growth inBiology. The interconnective pore size of foamed implants from mold Bhad a range of 91_6 mm to 67_7 mm。 by mold A this range was35_10 mm to 19_8 mm. This change was also corresponding to thefinding in the mean pore size of foamed implants from two molds. It could be concluded from Figures 3–6 that the improved mold designof mold B could not affect the change tendency of pore structure, such asdecreased pore size with rise of the injection speed, but it could increasethe porosity and the mean pore size as well as the interconnective poresize of the foamed implants. At the same time the