Meeting Banner
Abstract #2414

A Novel Tiered, Multi-Threaded Region Growing Algorithm for Improved Phase Correction for Two-Point Dixon Imaging

Jingfei Ma1, John D. Hazle1

1Imaging Physics, MD Anderson Cancer Center, Houston, TX, United States

We present a novel tiered, multi-threaded region growing algorithm for two-point Dixon water and fat imaging that can overcome the difficulty in phase correction when regions of large noise, artefacts, or isolated tissues are present. Each thread of processing starts from an initial seed and covers only a high-quality region containing the initial seed. Tiered seed masks serve as a quality metric and allow more trusted regions to be processed first even when completely isolated regions are present. Successful application of the algorithm is demonstrated for processing in vivo two-point Dixon images.

Keywords

accepted actual additionally advances afterwards already among amplitude anatomy angular another apparent applied assigned assignment automatic available background basic basis better body cancer challenging clinical complete completed complex consistent consistently context correct correction correctly correlation criterion dataset decided decision determination determined difficulty directly discarded dominant dual either equivalent equivalently even example exists experiment failure fast final finally fits forced fully fundamental gaps growing help identification identified implemented improved in vivo incorrect independent indeterminate initial isolated john known legs made mask masks metrics modifications modulation narrow neighborhood noise note noted novel orientation outside overcome parts pixel pixels previous primary priority proceed proceeds processed processing proposed pulse pure quality reconstructed recorded remaining requiring respectively scheme secondary seed seeded seeds selected selection separate separated serve several shelved simply slice smaller smooth spatial spatially start step steps still subjects swap technical temporarily tend tertiary thread threaded threshold tiered tissue trusted vector visit visited volume water whole whose