Machine Learning has been used increasingly in radiology and medical imaging because typical imaging objects such as lesions and organs presented in medical images are on most occasions far too complex to be represented reliably by a certain simple equation or hand-crafted model. Such models, and the simple features calculated from them, cannot effectively solve the medical imaging problem, or generally provide the discrimination power to reliably detect and classify objects of interest in individual patient images with variable indications. Solving the medical imaging problem using deep neural networks has re-gain attention due to the breakthrough of the GUP device and convolution neural networks architecture. Especially for image-to-image transformation, it has a high potential for quantification corrections and image prediction from different imaging modalities. In this talk, we will introduce our recent works on the application of neural networks for solving the medical imaging problem, such as dual isotope PET imaging, SPECT attenuation correction, metal artifact reduction in CT images, and PET imaging for proton range verification.
Zoom link:
https://ijclab.zoom.us/j/94392457912?pwd=RkZMUmdqRDJzTmw1Z0FkM3pjdSt6QT09
