This technology is a denoising device and method that applies Contrast-Limited Adaptive Histogram Equalization (CLAHE) to contrast-enhanced X-ray images, estimates the noise level of the resulting image, and performs Block-Matching 3D (BM3D) filtering based on that estimate. The noise level is determined by the standard deviation of blocks with the highest histogram similarity.
Histogram equalization, widely used for contrast enhancement in medical imaging, has historically suffered from the issue of amplifying noise along with contrast, leading to degraded image quality. Furthermore, it has been difficult to accurately determine the noise level in enhanced X-ray images, making it challenging to calibrate the intensity of subsequent noise reduction filters.
This technology accurately estimates noise standard deviation by decomposing CLAHE images into sub-images and blocks, then selecting homogeneous blocks using total gradients—derived from horizontal and vertical gradients—as a similarity metric. By applying this estimate to BM3D, it effectively removes noise while enhancing contrast. It can be integrated into angiography X-ray reading systems and medical imaging post-processing software, allowing radiologists to more clearly identify fine blood vessels and lesion boundaries.
N/A
