Introduction 1 2 3 5 6 7 8 9 10 11 12 13 1 14 17 Table 1 Overview of relevant literature Reference Gestational age (weeks) Scan week No. of patients No. with gestational age <32 weeks MR field strength (T) ROI 14 <37 28–43 27 <5 1.5 a 15 24–36 28–39 17 <5 1.5 Manual 16 24–36 28–39 14 <5 1.5 Manual 17 26–30 28–40 6 <5 1.5 Manual a Materials and methods Subjects The Erasmus MC Ethical Review Board approved the study and written informed parental consent was obtained for each subject. The images used in this study had been obtained within the framework of a study in which premature infants of different gestational ages underwent serial conventional and DTI acquisitions to evaluate white-matter development. The inclusion criteria for our study were birth at gestational age 25–32 weeks, no evidence of white-matter injury on conventional MRI and scanned within 4 days of birth. Developmental outcome assessed at age 1–3 years needed to be normal. Gestational age was calculated from the mother’s last menstrual period or estimated from early US (<18 weeks of pregnancy). Exclusion criteria were intraventricular haemorrhage, ventriculomegaly, congenital infection, brain malformation or a multiple congenital anomaly syndrome. Furthermore, preterm infants whose images showed severe motion degradation were excluded since evaluation of these scans was not possible. During the 25 months of the study period (March 2004 to April 2006), 41 infants were scanned and 32 met our study criteria. Four MR examinations were excluded due to severe motion artefacts, so 28 patients were included. Neurodevelopment monitoring 18 19 Conventional MR imaging 2 20 22 Diffusion tensor postprocessing 1 23 25 1 2 3 1 2 3 The tracts selected for quantization in the study included commissural tracts (corpus callosum: splenium and genu), projection tracts including those of the posterior limb of the internal capsule (PLIC), the anterior limb of the internal capsule (ALIC) and the optic radiation (OR), and association tracts (external capsule, EC). Regions of interest 26 1 2 3 4 5 Fig. 1 Example of an FA map (gestational age 28 weeks) on which the ROIs are placed  Fig. 2 Confirmation of ROI placement on the colour map (gestational age 28 weeks)  Fig. 3 Fibre tracking was performed in each of the ROIs to confirm the correct location. This is an example showing the corticospinal tracts (gestational age 28 weeks)  Fig. 4 Delineation of the corpus callosum using free-hand ROI placement on an FA map (gestational age 28 weeks)  Fig. 5 Automated calculation of the maximum FA pixel value within the ROI on an FA map (gestational age 28 weeks)  Statistical analysis The relationship between FA of the white matter tracts and gestational age was analysed by correlation analysis (Pearson product moment correlation, SPSS 13.0.1). Tract comparison was done using one-way analysis of variance (ANOVA). Results Patient characteristics Gestational ages at birth ranged from 26 to 32 weeks (mean 28 weeks 5 days). Mean weight at birth was 1,148 g and mean head circumference was 26.4 cm. All patients had shown normal neurodevelopment as defined by the Denver or Bayley scoring system according to the most recent examination. FA and ADC of ROIs of standard pixel size t P r P 6 2 2 Fig. 6 Average FA values of 16-pixel ROIs of the PLIC  Table 2 Tract statistics (FA and ADC mean values and standard deviations) and significant differences in FA and ADC between tracts Structure FA ADC Mean (SD) P Mean (SD) P Posterior limb of internal capsule 0.349 (0.028) a, b, c 1.09 (0.05) l, m, n, o, p Anterior limb of internal capsule 0.242 (0.033) a, d, e, f 1.27 (0.091) l External capsule 0.175 (0.188) b, g, h, i 1.29 (0.097) m Optic radiation 0.270 (0.042) g, j, k 1.30 (0.126) n Corpus callosum (genu) 0.42 (0.048) d, e, h, j 1.24 (0.111) o Corpus callosum (splenium) 0.442 (0.056) c, f, i, k 1.27 (0.131) p Maximum FA values of the PLIC r P 7 Fig. 7 Maximum FA ROI values of the PLIC  Comparison between tracts 2 Discussion 16 27 16 28 29 16 16 30 32 26 33 15 13 34 During development a decrease in ADC values of the white-matter tracts is expected. We found no correlation between average ADC values and gestational age in the studied tracts. When looking at the available data reported previously we only saw the ADC values decrease significantly with gestational age when the ADC values of the infants older then 35 weeks were included. The fact that we could not find any significant correlation could be due to the limited number of infants scanned. The alternative is that at an earlier gestational age there is no significant decrease in ADC in the studied tracts. Studying DTI parameters of the white-matter tracts of VLBW infants is challenging for many reasons. One of these challenges is to determine a standard for the size and shape of the ROIs. A ROI is a controlled identification of a given area of an image for numerical analysis and the area of anatomy being scanned that is of particular importance in the image. Different authors have used different ways to set their ROIs. The reason we compared two techniques was to achieve better reproducibility. Maximum values (the maximum pixel value within a ROI) might be an alternative in which the value is given by the software. However, theoretically there is a bigger change of artefacts. Fibre tracking, colour maps and ADC maps are established but time-consuming techniques for the verification of tracts. We trust that automatic verification will become common practice in the future. A possible solution for ROI comparisons between researchers is the use of a neonatal brain atlas coordinate system. Individual brain images could then be transformed into a common coordinate space and the ROIs could be placed at specific topographic coordinates. Our research group is currently looking into this option. Another serious challenge is the SNR and spatial resolution constraints due to the very low anisotropy of premyelinating white matter and the tiny size of white-matter tracts in premature newborns. Using a custom-made MR-compatible incubator with a high-sensitivity neonatal head coil that improved image quality, spatial resolution and patient comfort, we were able to overcome this challenge. 2 35 Like all VLBW infants discharged from our NICU, our study patients were routinely seen by trained paediatricians and by paediatric physical therapists at the outpatient department for neurodevelopmental follow-up. Neurodevelopmental outcome was defined according to the most recent neurological examination and Denver and Bayley scores. The children’s ages at the most recent assessments varied, making the results difficult to compare; also Denver and Bayley scores are limited in their prognostic value below the age of 2 years. Better understanding of normal preterm white matter development is essential to encourage the use of DTI for evaluation and treatment of white-matter injury. Early diagnosis of white-matter abnormalities means that early intervention might be possible. We are exploring the feasibility of perinatal brain repair, and new MR imaging techniques such as DTI will enable us to improve our understanding of how the developing brain responds to our interventions. Conclusion Our study gives anisotropy values for VLBW infants with normal outcome that can be used as reference values. This work adds to our understanding of normal preterm white-matter development.