brain volume estimation from post-mortem newborn and fetal mri

13
Brain Volume Estimation from Post-Mortem Newborn and Fetal MRI Department of Forensic and Medicolegal Medicine Faculty of Medicine University of Syiah Kuala February, 2015 Putri Rahmi Farah Dina F Kazi Muntazeri Cut Almira Honesta Novita Yuniwanda I

Upload: farah-firmandez

Post on 12-Sep-2015

8 views

Category:

Documents


2 download

DESCRIPTION

forensik

TRANSCRIPT

Brain Volume Estimation from Post-Mortem Newborn and Fetal MRI

Brain Volume Estimation from Post-Mortem Newborn and Fetal MRIDepartment of Forensic and Medicolegal Medicine Faculty of MedicineUniversity of Syiah KualaFebruary, 2015

Putri Rahmi Farah Dina F Kazi Muntazeri Cut Almira Honesta Novita Yuniwanda IBrain Volume Estimation from Post-Mortem Newborn and Fetal MRIEliza Orasanua,, Andrew Melbournea, M. Jorge Cardosoa, MarcModata, Andrew M. Taylorb, Sudhin Thayyilc, Sebastien Ourselina

www.elsevier.com vol 6 September 4, 2014BackgroundThe loss of a fetus, a baby or a child is traumatizing and devastating for a parent. Over the past decade, the consent rate for autopsy in the newborns has been less than 20% and less than 50% in stillborns in the United KingdomMRI is a powerful tool that can be used as a post-mortem imagingtechnique with high accuracy and high level of performance for depictingsoft-tissue lesionsAccurate estimation of brain weight is an integral part of autopsy, since any deviation from the normal ranges could be an indicator of pathological change in the organ and therefore helps in establishing the cause of death.MethodsMethods

Cerebrum + CSF CerebellumBrainstemNewborn0.991 0.002 0.873 0.028 0.819 0.031Fetus0.992 0.002 0.888 0.016 0.815 0.035Fetus (public atlas) 0.984 0.012 0.756 0.185 0.676 0.262Dice overlap of automatic segmentation with manual segmentation for each structure among each cohortTable 1MethodsResultAtlas-based brain segmentationPost-mortem cerebrum volumesTable 2Comparison between MRIweights,MRImanualweights and autopsyweights of the cerebrum for newborns. Density for cerebrumused to compute the weight from the MR volume 1.08 g/mL. The threshold values used to remove the CSF from the MR segmentation are also included.MaRIAS no.Gestational age [weeks]Postnatal age [days]CSF threshold value MRI volume [mL] MRI weight [g] MRI manual weight [g] Autopsy weight [g]19340 21 600418.15 451.61 446.08 457.0194391600418.56452.05446.46462.5256 38 1 800 360.43 389.27N/A 433.025929120 400413.45446.52442.92 456.7262 373500338.13365.18N/A 368.0274402600430.50481.29N/AN/A276 411600333.32359.99N/A 367.02933510600329.07355.40 350.62360.0300401400388.61419.70N/A420.0306371500287.15345.49309.86N/A313373800310.26335.08 N/A360.0315403 900321.41347.13347.30409.031840 2400313.07]338.12]N/A368.03254030500586.24633.14627.79635.0356401 600428.28462.54N/A445.9368355600345.69373.35N/A 363.83864030700501.73541.87534.37609.0Table 3Comparison between MRIweights,MRImanual weights and autopsyweights of the cerebrum for fetuses. Density for cerebrumused to compute the weight fromtheMR volume 1. 08 g/mL. The threshold values used to remove the CSF from the MR segmentation are also includedMaRIAS no.Gestational age [weeks]Gestational age [days]

CSF threshold value MRI volume [mL] MRI weight [g] MRI manual weight [g] Autopsy weight [g]12139 0700361.74390.67390.40388.00149342700267.79289.22N/AN/A19632 2800160.66173.52 172.59181.00210386600354.74383.12N/A391.00214372550 298.07321.92N/A 328.00230416450370.73400.39N/A401.00246 326530208.06224.70225.54223.70282305500146.96 158.72N/AN/A299440300408.43441.11438.00447.00330356800264.72285.90N/A 292.00339 362800339.31366.45 ]365.05N/A343 341350247.69267.50N/A275.00384394300366.63395.96N/A 389.80384394300366.63395.96N/A 389.80388380390439.41474.56469.56N/A391 0 330550213.39230.46231.59226.00396290700113.70122.80122.22128.00ConclutionWe segmented the cerebrum and CSF, cerebellum and brainstem of post-mortem newborn and fetal subjects of MaRIAS using atlas-based segmentation and an EM-based algorithm with relaxation priors

We validated the semi-automatic segmentation by computing the Dice overlap with manual segmentations

The excellent agreement of the segmentationderived weights with the autopsy ones are of great importance, since it supports the use of post-mortem MRI, currently alongside otherminimally invasive procedures, as a viable alternative fo conventional autopsy of infants and fetusesTHANK YOU