Profile
Visiting Professor
Professor KUDO Hiroyuki
Institute of Systems and Information Engineering, The University of Tsukuba
- Themes
-
Computed Tomography (CT), Imaging Science, Deep Learning, Image Processing, Medical Imaging, Inverse Problems
- Keywords
- Computed Tomography (CT), Imaging Science, Deep Learning, Image Processing, Medical Imaging, Inverse Problems
- Research Activities
-
Research on Computed Tomography (CT) and Imaging Science
*Research on Computed Tomography (CT) For a long time, we are continuing research on Computed Tomography (CT), which aims at visualizing internal structures of objects in the form of cross-sectional images by a non-destructive way. In CT, we measure data called projection data by using various quantum beams such as X-rays, Gamma-rays, and electron beams, and cross-sectional images are generated from the projection data by performing data processing called "image reconstruction". It can be said that the image reconstruction is a key of CT technologies. In the past, we have performed a variety of researches on image reconstruction for medical X-ray CT, Positron Emission Tomography (PET), Electron Tomography, X-ray phase CT, and Synchro Radiation CT. In more detail, we have developed 1) image reconstruction methods for 3-D cone-beam CT, 2) image reconstruction methods to generate high-quality images from sparsely measured or low-dose projection data to reduce radiation dose and data acquisition time, and 3) image reconstruction methods in interior CT which radiates X-rays only to a small Region-of-interest (ROI) such as heart or breast to reduce patient dose. As a member of SRIS, we are now performing research on image reconstruction for new CT named as "multi-beam CT", which is able to generate 4-D (spatial + time) images with 1 (ms) temporal resolution. As a member of SRIS, I will aim at using the multi-beam CT to elucidate physical phenomena for which the physical mechanism is still unknown.*Research on Imaging Science In the fields of image processing which treats image restoration and image analysis, the following dramatical changes occurred around 2000. Methodologies such as mathematical optimization and compressed sensing had been introduced into the image processing, and its performance and accuracy had been significantly improved compared to the classical methodologies. Since then, this new field has been called by the new name "imaging science". We are working on developments of a variety of image processing algorithms based on the concept of imaging science with applications to medical imaging as well as natural image processing.*Research on Deep Learning Image Processing Thanks to the wide spread of Deep Learning (DL), the way to develop image processing algorithms has been changed from the classical hand-crafted approach to the data-driven approach using DL. This can be considered a very big break-through in the field of image processing. We are working on the development of a variety of image processing algorithms for image reconstruction in CT, image restoration, and image analysis based on DL.
Schematic illustration of research contents of our laboratory.
Example reconstructed image of multi-beam CT equipment by a newly developed image reconstruction method. The sample object is a glass capillary.
- Message
- I will actively conduct research as a member of SRIS by based on my past achievements on data processing methodologies for imaging equipment.
