Diffuse lung disease (DLD) is difficult to diagnose due to the ambiguity of disease patterns, which motivates the development of image retrieval method to facilitate the physicians in diagnosis by retrieving the similar cases from database. In this paper, we propose a similarity measurement method for diffuse lung disease computed tomography (CT) slice image retrieval. In our method, the DLD patterns and the spatial distribution of the diseased area are both integrated to compute the similarity between query and database image. For this purpose, the powerful GoogLeNet is adopted and fine-tuned to locate the diseased area and classify it into different DLD patterns. Moreover, the spatial distribution of the diseased area is calculated based on the distance to the body center. Our method is verified on 324 CT slice images obtained from 53 subjects. The correct ratio among the top-5 retrieved images achieved 86.2%. Based on this performance, we can draw the conclusion that this method has high potential to improve the efficiency for diagnosis of diffuse lung disease in clinical use.