A volume filtering and rendering system for an improved visual balance of feature preservation and noise suppression in medical imaging
Preserving or enhancing salient features whilst effectively suppressing noise-derived artifacts and extraneous detail have been two consistent yet competing objectives in volumetric medical image processing. Illustrative techniques (and methods inspired by them) can help to enhance and, if desired, isolate the depiction of speciﬁc regions of interest whilst retaining overall context. However, highlighting or enhancing speciﬁc features can have the undesirable side-effect of highlighting noise. Second-derivative based methods can be employed effectively in both the rendering and volume ﬁltering stages of a visualisation pipeline to enhance the depiction of feature detail whilst minimising noise-based artifacts. We develop a new 3D anisotropic-diffusion PDE for an improved balance of feature-retention and noise reduction; furthermore, we present a feature-enhancing visualisation pipeline that can be applied to multiple modalities and has been shown to be particularly effective in the context of 3D ultrasound.