Application of a neural network to automatic gray level adjustment for medical images

The authors have developed a system to automatically adjust the gray level of magnetic resonance (MR) images using a neural network. The gray level of an MR image is adjusted by setting the display window (gray-level) width and level (WWL). The authors define an index, EW, for the evaluation of displayed image clarity, and they prove its effectiveness. They use a neural network to learn the relationship between image histogram features and displayed image clarity. The authors calculated image clarity using the NN, performed two-stage searching, and determined the best possible WWL. They also evaluated the WWL adjusted by the system using the clarity index, EW.<
>