Low dose CT image restoration using a localized patch database

As growing concerns on the potential side effect of radiation induced genetic, cancerous and other diseases, the low-dose CT imaging is becoming a hot issue, how to minimize the radiation exposure level while maintaining the diagnostic performance. In previous work, we have proposed a framework that is working in conjunction with image-based database. It restores a low-dose image with modified non-local means (NLM) that have expanded search window to include regular-dose priors obtained from same or different patients. In this paper, we further develop the framework to know the internal structures of CT images so that the NLM filtering is applied with more intelligent way. In addition, knowing the internal structures changes the database from image-based to patch-based that organized structure by structure. As preliminary work, this was tested to restore a structure of low-dose CT image and it showed a very promising results.