Abstract：
The concentration of the polymer in the electrospinning solution greatly influences the mechanical behaviour of electrospun vascular grafts due to the influence on scaffold morphology. The scaffold morphology (fiber diameter, fiber orientation and inter-fiber voids) of the grafts plays an important role in their behaviour during use. Even though manual methods and complex algorithms have been used so far for characterisation of the morphology of electrospun architecture, they still have several drawbacks that limit their reliability. This study therefore uses conventional, statistical region merging and a hybrid image segmentation algorithm, to characterise the morphology of the electrospun vascular grafts. Consequently, vascular grafts were fabricated using an in-house electrospinning equipment using three polymer material concentration levels (14%, 16% and 18%) of medical-grade thermoplastic polyurethane (Pellethane®). The image thresholding and segementation algorithms were then used for segmentation of SEM images extracted from the polymer grafts and then morphological parameters were investigated in terms of fiber diameter, fiber orientation, and interfiber spaces (pore area and porosity). The results indicate that electrospun image segmentation was \"best\" when the hybrid algorithm and the conventional algorithm was used, which implied that fiber property values computed from the hybrid algorithm were closed to the manually measurements especially for the 14% PU with fiber diameter 2.2%, fiber orientation 7.6% and porosity at 1.9%. However there was higher disperity between the manual and hybrid algorithm. This suggests more fiber uniformity in the 14%PU potentially affected the accuracy of the hybrid algorithm.
摘要：
由于对支架形态的影响,电纺溶液中聚合物的浓度极大地影响电纺血管移植物的机械行为。支架形态(纤维直径，移植物的纤维取向和纤维间空隙）在使用过程中对其行为起着重要作用。即使手动方法和复杂的算法已被用于表征静电纺丝结构的形态，到目前为止，它们仍然有几个缺点限制了它们的可靠性。因此，这项研究使用传统的，统计区域合并和混合图像分割算法，表征电纺丝血管移植物的形态。因此,使用内部静电纺丝设备，使用三种聚合物材料浓度水平（14%，16%和18%)的医用级热塑性聚氨酯(Pellethane®)。然后使用图像阈值和分割算法对从聚合物移植物中提取的SEM图像进行分割，然后根据纤维直径研究形态学参数，纤维取向，和纤维间空间（孔面积和孔隙率）。结果表明，当使用混合算法和常规算法时，电纺图像分割是“最好的”。这意味着从混合算法计算的纤维属性值与手动测量接近，特别是对于纤维直径为2.2％的14％PU，纤维取向7.6%，孔隙率1.9%。然而，手动算法和混合算法之间存在更高的分散性。这表明14%PU中更多的纤维均匀性潜在地影响混合算法的准确性。
