This paper proposes a new algorithm by improving and combining S-DES and Back Propagation in neural networks to identify the stego image in mails. Back propogation is ideal for simple pattern recognition and mapping tasks. This new algorithm checks the mail inbox for JPEG files. As the image is send across the network for transmission, compression is required. Compression is used to minimize the seeking time of the JPEG files. After compression, the files are stored in a large database for further processing. A sample image is taken from the database as cover channel which is used to hide the secret information. For each image, we embedded a random binary stream of different lengths using S-DES algorithm. The proposed research analyses the performance of the improved version of image steganalysis algorithms in corporate mails.