Coronary angiogram image enhancement using decimation-free directional filter banks

The detection and enhancement of coronary arterial trees (CATs) in an angiogram image is an important preprocessing task that greatly reduces the stress on further processing such as 3D reconstruction of a CAT model. Conventional techniques make use of gradient operators to detect the CAT structure. However, the gradients are local operators that do not provide a continuous map of arterial trees, especially in a noisy environment. We propose a decimation-free directional filter bank (DFB) structure. It provides an output in the form of directional images as opposed to directional sub-bands provided in previous DFBs. The presence of directional images facilitates any further spatial processing if needed. However, we have to prepare an angiogram image before it can be given as input to the proposed DFB structure due to the fact that acquired angiograms are low in contrast. The preparation steps involve removing non-uniform illumination from the image. Then the DFB structure outputs directional images. The final enhanced result is constructed on a block-by-block basis by comparing the energy of all the directional images and picking one that provides maximum energy. The enhancement that results in the final image is due to the fact that we can separate omnidirectional background noise from the CAT structure which is predominantly a directional feature.