| 摘要: | 積層製造的技術近年來廣泛應用於生物支架的製作，以建構出具有特定孔隙率的支架。支架上的細胞能藉由支架間的孔洞，有效的與外界進行養分及代謝物的交換，促進細胞生長。為了製作具有穩度孔隙率的支架，必須精準的控制支架的線徑寬度，若支架的線徑寬度不均，將導致孔洞大小不一，使組織液在支架的孔隙中流通不易，無法有效的與細胞進行養分交換，且不穩定的線徑，會降低支架整體的機械性能，造成外型坍塌的風險。
水凝膠作為支架的主流材料之一，其特性為材料黏度會隨著溫度的改變而發生變化。因此使用水凝膠做為生物墨水列印時，墨水成形方式若採用溫度感應成型機制，需要精確的掌控沉積面溫度。傳統上大多採用單點接觸式溫度感測器量測墨水沉積面溫度，缺點為無法完整呈現同一層的溫度分佈。在同一層高度，隨著量測位置的不同，得到的溫度也會改變。且量測時為了不要接觸到支架造成溫度變化，會將感測器放置於沉積面上，導致量測的溫度與實際沉積面溫度產生偏差。
本研究之目的為預測支架在不同沉積溫度以及列印參數下的線徑寬度。為獲取溫度 資料，會使用熱像儀拍攝支架預沉積面之二維熱像圖，可快速有效獲得位置溫度資料。 搭配自行建立之深度神經網路訓練，以溫度與列印參數設定值(材料擠出壓力、噴嘴移動 速度)作為輸入參數，支架的線徑寬度作為輸出，預測溫度和列印參數與支架線徑的關係。 本實驗總共製作了 348 筆數據，298 筆用於訓練集，50 筆用於驗證集，R 2 為 98.5%， 證實模型學習的成果具有可信度。利用模型的預測能力，未來可控制生物列印機，在每 一個列印點區間調控噴嘴移動速度，以獲得連續一致的線徑寬度，達到提升支架列印品 質的目的。
;Nowadays, the technology of Additive Manufacturing(AM) has been widely applied to the fabrication of biological scaffolds to construct scaffolds with specific porosity. The cells on the scaffold can exchange nutrients and metabolites effectively with the outside environment through the pores between the scaffolds to promote cell growth. In order to make scaffolds with stable porosity, it is necessary to precisely control the linewidth of the scaffolds. If the linewidth of the scaffolds is non-uniform, the pore size will be different, which makes it difficult for the tissue fluid to flow in the pores of the scaffolds and cannot effectively exchange nutrients with cells. Besides, unstable linewidths will reduce the overall mechanical properties of the scaffolds and cause scaffolds collapse.
As one of the mainstream materials for scaffolds, hydrogel is characterized by changes in material viscosity with changes in temperature. Therefore, when using hydrogel as a bio-ink for printing, the ink forming method requires precise control of the deposition surface temperature if a temperature-sensitive forming mechanism is used. Normally, single-point contact temperature sensor is used to measure the temperature of the ink deposition surface, which has the disadvantage of not being able to accurately present the exact temperature distribution of the same layer. In the same layer height, with the different measurement position, the temperature obtained will also change, and in order not to touch scaffolds to cause temperature changes, the sensor will be placed above the deposition surface, resulting in the measured temperature and the actual temperature of the deposition surface deviation.
The purpose of this study is to predict the linewidth of the scaffold at different deposition temperatures and printing parameters. To obtain temperature data, 2D thermal images of the pre-deposited surface of the scaffolds is taken using a thermal imaging camera. The temperature and print parameters (pressure, speed) were used as inputs and the linewidth of the scaffolds is used as output to predict temperature and printing parameters versus linewidth. In this study, a total of 348 data were produced, 298 for the training set and 50 for the validation set, with an R 2 of 98.5%, confirming the reliability of the model learning results. With the predictive capability of the model, the bio-printing can be controlled in the future to regulate the nozzle movement speed at each printing point to obtain a continuous and uniform linewidth to improve the scaffolds quality. |
