Size measurement of nanoliposomes
Mean diameter of nanoliposomal artemisinin was measured 500 nm and that of pegylated nanoliposomal artemisinin was 455 nm which was considered to be in nano dimension size.
Encapsulation efficiency
Encapsulation efficiency (EE %) was calculated through spectrophotometry method. Regarding encapsulation efficiency formulation, the amount of nonencapsulated artemisinin is obtained, which leads to determination of the amount of encapsulated drug, considering the amount of the initial drug. Encapsulation efficiency is 96.02% for nanoliposomal artemisinin formulation and 91.62% for pegylated nanoliposomal artemisinin formulation.
Toxicity
MTT technique was hired to obtain toxicity. IC50 for nanoliposomal, pegylated nanoliposomal and standard drug is 2.1, 1.58 and 2.7 μl/ml by means of Pharm program.
Artificial neural networks
Artificial neural networks are characterized by network architecture, node characteristics, and learning rules. The operation of neural networks is divided into two stages: learning (training) and generalization (recalling). Network training is typically accomplished by using examples, and network parameters are adapted using a learning method. This can be done in an online or offline manner. Once the network is trained to accomplish the desired performance, the learning process is terminated and it can then be used directly to replace the complex system dynamics. The trained network can be used to operate in a static manner: to simulate an unknown dynamics or nonlinear relationship Arbib (2002). Neural networks are usually biologically motivated. Each neuron is a computational node, which represents a nonlinear function. Most applications of neural networks utilize the function approximation capability of neural networks. These include modeling and system identification, regression and prediction. The neural networks are trained using supervised learning Feng and Ratnam (2000). The input and output layers are defined according to the dimensions of the input and output patterns. The input–output pattern pairs are fed to the network, and network parameters are adjusted accordingly. A trained network represents the learned nonlinear relation between the input–output pairs, and can generalize when an unknown input pattern is presented to the network Tipping (2001).
Artificial neural network modeling
In this study, cytotoxicity of artemisinin in breast cancer treatment by means of feed-forward artificial neural networks is modeled and predicted. Thus, a three-layer neural network is designed which incorporates 6 neurons in hidden layer and one in input layer for drug concentration and one neuron in output layer for cytotoxicity. Transfer function of Radbas is used for hidden layer neurons and Purlin transfer function is used for output layer; and Levevberg-marquardt backpropagation algorithm was used as network training procedure. Then, four sets of data including drug concentration and cytotoxicity are implemented to the network. After the training stage, two sets of new data, which have had no roles in network training, are implemented to the network as input according to Figure 1. And then, network output, including cytotoxicity, was anticipated.
Figure 2 and Figure 3 reveal that neural network anticipation and experimental values for cytotoxicity are closely similar; in which network anticipation for cytotoxicity of nanoliposomal drug correlation of coefficient is R = 0.9104 with a mean square error of 29.3837 and that of pegylated nanoliposomal drug correlation of coefficient is R = 0.8925 with a mean square error of 37.0219.