A content-based medical teaching file assistant for CT lung image retrieval

In this paper, a content-based scheme for assisting the construction of a teaching file system to retrieve lung Computed Tomographic (CT) images is presented. The system uses visual-based user interface to allow the user to enter or query an image by selecting the region of interest (ROI); and uses neural network to classify the relationship between the images stored in database. The system will output a set of candidate images that are textural-similar to the query image. We marked the abnormal portions of each training image by rectangular shape manually because it needs the knowledge of expertise. Then, the texture features of each marked region are extracted by selecting the most important coefficients of 2D FFT. In the training stage, the system uses a Kohonen self-organizing network to classify those extracted FFT coefficients. In the query stage, the system first checks which texture category the query image in, then uses some geometrical characteristics to identify the most likely candidate image. The experimental results show that on average 92% of original images can be correctly retrieved with the displacement up to 22% of the block size