Comparative analysis of locally adaptive image enhancement for 3D ultrasound images

In this article we compare methods for locally adaptive preprocessing of 3D ultrasound (US) images and their different capabilities of enhancing diagnostically important anatomical structures. We show that suitability and optimization of methods depend on the objective of the preprocessing task, e.g., slice representation or volume rendering. This article analyzes which features should be used to control the enhancement algorithm For slice representations optimization of local statistics is appropriate to maintain diagnostically important structures. Features with good region separation properties are preferable for volume rendering tasks. We demonstrate our results with US images from clinical examinations.