The reduction of image file size by means of lossy compression algorithms such as JPEG 2000 can lead to distortions in image content affecting the detection of structures of interest. For microscopic images of blood cells, which usually occupy large amounts of disk space, the use of lossy compression is justified within a controlled quality loss. The present paper deals with a strategy for the automatic compression rate estimation where maximum reduction of file size is achieved while image quality is preserved. The compression rate is estimated from the analysis of Structural Similarity index metric behavior, instead of its statistical value, and its correlation with subjective tests performed by human observers.