Meeting Banner
Abstract #3171

DTI in Neonates: Data Corruption Due to Motion

Anneriet M. Heemskerk1, 2, Annemarie Plaisier2, Irwin Reiss2, Maarten H. Lequin1, Alexander Leemans3, Jeroen Dudink, 12

1Radiology, Erasmus MC University Medical Center, Rotterdam, Netherlands; 2Pediatrics, Erasmus MC University Medical Center, Rotterdam, Netherlands; 3Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands

Motion corrupted slices are a major problem in neonatal DTI as neonates have high occurrence of movement and this motion results in corrupted slices. We investigate the number of outliers in the tensor estimation as a method to detect corrupted slices in neonates. We found that 60% of the 27 subjects had corrupted data, defined as >10 slices with >30% outliers. These corrupted slices can lead to registration problems and erroneous tensor estimates depending on their location and spread. Targeted acquisition and processing is needed in this specific subject group to obtain reliable tensor estimates.

Abstract PDF E-Poster Video

Keywords

acquisition additional additionally advanced affected affine analyzed applied approved artifacts assessment board born care careful carefully child clinical compatible concern consistent containing corrected correction corrupted corrupting corruption datasets depends detect determine determined differently diffusion dramatically drop eddy effectiveness erroneous especially estimation ethical evaluated even evidence expect expectancy extra fraction gestational give gradient handled head heavily impact incidence inclusion incubator indication injury institute insufficient intensity intra investigate investigated larger lead least limited local location longer loss major maps material matrix measures medical miscalculation morris motion movement need needed neonatal neonates objective occurrence ongoing ordinary outliers pediatrics placed population presses problem problems processing producing profile prospectively quality quantified quantify radiology registration rejection related reliability reliable report reported restore review routine scanned scanning sciences sedated several severe shake shears short slice slices squares stable stretches studies subject subjects subpopulation sudden table targeted tensor thereby tools turn typically underlying underwent unsettled vendors week weeks white