Perceiving the connectivity between different regions of human’s brain is considered as an important objective in Neuroscience. Modelling of connectivity and how the functional interaction influence different brain regions has profound impact on medical and neuroscience fields. There are several methods for recording the brain signals and analyzing the neural activity of regions, in the Neuroscience. functional MRI is identified as one kind of well-known methods which records the brain signal. Because of the appropriate spatial and temporal resolution of recorded data, these data has been investigated and used effectively by scientists and doctors. Analyzing the recorded signal by functional MRI and recognizing the connectivity of brain regions is taking place to two general methods of data analysis or model analysis. Less complexity and great accuracy are almost always considered as two important principles in neuroscience matters and data evaluation of brain activity. Combination of data and model analysis methods are considered more reliable. During the recording data, some regions of brain are active. Identification and separation of these active regions have remarkable influence on effectiveness of the model. To achieve the proper model, scientists assume some general linear models. The recorded data are the output of models, and subsequently they try to estimate the parameters of the model correctly. Design matrix and its weight are among the most important parameters to estimate. The columns of the design matrix are specifically the weights of the model. The estimated parameters after statistical examinations present the results of active regions. The design matrix and methods have significant influence on accuracy of results. In our proposed method, according to improved general linear model, a flexible design matrix based on data and controlling commands is presented. Limitations of getting real functional MRI data, the online databases and some synthetic data were utilized. The active regions were specified more accurately using this method. Based on the proposed model, active brain parts are determined using special controlling commands with the accuracy of 96%. Key words: 1. functional MRI , 2. Bilinear Deterministic Dynamic Causal Modelling , 3. General Linear Modelling , 4. Design Matrix