题 目:Monogenic signal theory for radiological images

时 间: 2016年9月29日(周四)10:20

地 点:数统学院203会议室

报告人: Professor Uwe Kähler （Department of Mathematics,University of Aveiro, Portugal）

摘 要: Medical ultrasonography imaging for nodule detection is a non-invasive diagnostic test, which combines low cost, short acquisition time, and sensitivity to the number and size of abnormal nodules. However, a chief problem is that ultrasound images have low contrast, making it hard for the experts to interpret and classify the nodules detected. In this talk we discuss techniques based on the concepts of monogenic signal which aims to enhance the edges of abnormalities.

After discussing the correct mathematical definition of “edges” and “curves” in images we are going to use a combination of Riesz transforms and monogenic curvelets in order to determine the phase and phase angle of a given image. Using the remarkable properties of Riesz transforms: they are shift- and scale-invariant, preserve L^2 inner-product, and are steerable we are going to show how to determine size and position of abnormalities present in images.