Feature map guided medical image enhancement using (HE)lut mapping with anchor point group

Low contrast images in medical x-ray imaging would interfere with the regular diagnosis. Thus, contrast enhancement is of pivotal significance in addressing this problem. Multiscale enhancement method enables fine detail enhancement as well as visually appealing images. However, it may not control the lut (look up table) mapping function effectively, which may cause loss of weak signal or produce washed-out appearance. We propose a novel method called histogram equalization (HE) lut mapping with anchor point group (LAPG) to control the lut mapping function precisely, using 3 anchor points at least. Then a half automatic method is designed to obtain the anchor points' ideal values, which is guided by a feature map based on tissue segmentation. At last, the method of lut mapping with anchor point group is extended to the multiscale framework for advanced contrast enhancement. The experimental results, compared with single point or without any anchor point, show that anchor point group mapping could prevent information loss or saturation, and the half automatic anchor point decision method could save much time for parameters adjustment.