Background: brain sonography(U.S) is currently the reference screening method for early detection of periventricular leukomalacia (PVL) and periventricular haemorrhage (PVH).During the neonatal period, MRI assesment of brain damage is difficult because of brain immaturity (unmyelinisation and brain water content). Aims: 1) to achieve a quantitative T2 MRI map** (T2 map) of brain damage in premature infants during the neonatal period 2) to compare U.S, qualitative MRI (T2 weighted) and T2map findings to outcome at 18 mo of age. Methods:20 preterm babies, 26-33 weeks of gestation (mean 29) were studied at 37 wks conceptionnal age. MR scans were performed with a Brucker 2.35 Tesla. Spin-echo T2-weighted transverse scans were acquired using 3000ms/30,60,90,120,150 ms/2(TR/TE/Nex). Qualitative analysis of the MRI were performed on the fifth echo(TE=50ms) and on the T2map. Neurologic outcome was assesed using the Griffith test. Results:the table lists the T2map values in the periventricular white matter (w.m).For babies with good outcome, T2map increased from occipital to frontal region (normal myelination progress). For the babies with bad outcome T2map values were higher in the parietal and occipital regions.

Table 1
Full size table

Except in frontal w.m T2map allowed better detection of w.m lesions: Sensibility: U.S: 72%; T2 weighted: 82%; T2map:100%. Specificity: US:75%; T2 weighted: 75%; T2map: 75%. Positive predictive value: U.S: 80%; T2weighted 82%; T2map: 82.5%. Negative predictive value: U.S: 54.5%; T2weighted: 75%; T2map: 100% Conclusion: Quantitative T2 imaging leads to a better detection of parietal-occipital white matter damage and increases the value of MRI to early (near term) detection of brain injury.