Volume of hippocampal subfields and episodic memory in childhood and adolescence

NeuroImage 94 (2014) 162–171

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

Volume of hippocampal subfields and episodic memory in childhood and adolescence Joshua K. Lee a,b,⁎, Arne D. Ekstrom a,c, Simona Ghetti a,b a b c

Department of Psychology, University of California, Davis, 135 Young Hall, One Shields Avenue, Davis, CA 95616, USA Center for Mind and Brain, University of California, Davis, 202 Cousteau Place, Davis, CA 95618, USA Center for Neuroscience, University of California, Davis, 1544 Newton Court, Davis, CA 95618, USA

a r t i c l e

i n f o

Article history: Accepted 8 March 2014 Available online 15 March 2014 Keywords: Development Memory Hippocampus Dentate gyrus CA1 CA3

a b s t r a c t Episodic memory critically depends on the hippocampus to bind the features of an experience into memory. Episodic memory develops in childhood and adolescence, and hippocampal changes during this period may contribute to this development. Little is known, however, about how the hippocampus contributes to episodic memory development. The hippocampus is comprised of several cytoarchitectural subfields with functional significance for episodic memory. However, hippocampal subfields have not been assessed in vivo during child development, nor has their relation with episodic memory been assessed during this period. In the present study, high-resolution T2-weighted images of the hippocampus were acquired in 39 children and adolescents aged 8 to 14 years (M = 11.30, SD = 2.38), and hippocampal subfields were segmented using a protocol previously validated in adult populations. We first validated the method in children and adolescents and examined age-related differences in hippocampal subfields and correlations between subfield volumes and episodic memory. Significant age-related increases in the subfield volume were observed into early adolescence in the right CA3/DG and CA1. The right CA3/DG subfield volumes were positively correlated with accurate episodic memory for item–color relations, and the right CA3/DG and subiculum were negatively correlated with item false alarm rates. Subfield development appears to follow a protracted developmental trajectory, and likely plays a pivotal role in episodic memory development. © 2014 Elsevier Inc. All rights reserved.

Introduction Episodic memory is the capacity to remember details about events and critically depends on the hippocampus which integrates the arbitrary features of an experience into memory (Eichenbaum and Cohen, 2001). Episodic memory develops during childhood and adolescence (Ghetti and Lee, 2011; Ofen and Shing, 2013). However, the contribution of the hippocampus to this development has received little attention. Studies of age differences in the hippocampal volume have yielded contrasting results. Although age-related volumetric increases have been reported from childhood into young adulthood (Østby et al., 2009), other studies failed to find age differences (e.g., Giedd et al., 1996; Yurgelun-Todd et al., 2003). However, the trajectory of development may differ along the longitudinal axis of the hippocampus, even while the overall volume is relatively stable (DeMaster et al., 2013; Gogtay et al., 2006). Using different volumetric methods, DeMaster et al. (2013) and Gogtay et al. (2006) showed that with age, the hippocampal head decreased in volume while the hippocampal body increased; ⁎ Corresponding author at: 202 Cousteau Place, Davis, CA 95618, USA. E-mail address: [email protected] (J.K. Lee).

http://dx.doi.org/10.1016/j.neuroimage.2014.03.019 1053-8119/© 2014 Elsevier Inc. All rights reserved.

further, associations between volumes and episodic memory differed as a function of age and sub-region. These results raise the question of whether differences along the longitudinal axis reflect heterogeneity in hippocampal cytoarchitecture. The hippocampal formation comprises several subfields (Insausti and Amaral, 2004; Insausti, 2010), including the dentate gyrus (DG), the cornu ammonis (CA) subfields CA3, and CA1, as well as the subicular complex. Since these subfields are not uniformly distributed along the longitudinal axis, age-related differences in the subfields may help account for previous results. To date no published study has examined age differences in the volume of hippocampal subfields and their contribution to episodic memory in children. The present study begins to address this gap by pursuing three goals. First, we sought to validate with children a subfield segmentation method previously used with adults (Ekstrom et al., 2009; Zeineh et al., 2001, 2003). Second, we sought to conduct an initial investigation of age-related differences in the volume of hippocampal subfields in the body of the hippocampus, given that most protocols developed for 3 T imaging methods yield reliable segmentation of the subfields restricted to the hippocampal body. We expected volumes of the CA1 and CA3/DG, but not the subiculum, to increase with age, consistent with the results of the only available human development subfield data coming from a post-mortem

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

investigation (Insausti, 2010), as well as with results from developing macaque primates (Jabès et al., 2010). Furthermore, DG may exhibit protracted development possibly due to neurogenesis (e.g., Kalkan et al., 2013; Jabès et al., 2010; Lavenex and Lavenex, 2013; Yu et al., 2013) and myelination (Abrahám et al., 2010). Heterogeneous development of the subfields in the hippocampal body would provide initial support for the hypothesis that such development could account for age differences along the longitudinal axis reported previously (e.g., DeMaster et al., 2013; Gogtay et al., 2006). Third, we sought to explore the relation between subfield volumes and episodic memory in children. Evidence from adult humans (Kirwan and Stark, 2007; Shing et al., 2011; Yassa et al., 2011) and rodents (e.g., Hunsaker and Kesner, 2013; Sahay et al., 2011) suggests that the DG and CA3 may support encoding and retrieval of distinct event memories resistant to over-generalization (e.g., Leutgeb et al., 2007; Stark et al., 2013). Further, the protracted neurogenesis in the DG makes this field a good candidate for capturing the association between hippocampal volume and episodic memory during development. Thus, we predicted that episodic memory performance would be positively associated with CA3/DG volumes. This relationship is predicted to persist after statistically accounting for age. Materials and methods Participants Thirty-nine children (19 girls) participated in the study (M = 11.30 years, SD = 2.38, range 8 to 14 years). One additional participant was excluded from analysis because we incidentally discovered a brain abnormality of clinical significance (age 9). Participants and their families were recruited from the Davis and Sacramento areas and included mostly middle class families (family income, M = 87 K, SD = 28 K). Participants were not eligible if left-handed, had a psychiatric diagnosis via parental report (e.g., ADHD, dyslexia, depression), history of head trauma, premature birth (b 36 weeks), low birth weight (b5.5 lb), color-blindness, or any factor that related to participant safety in MR imaging environments. Informed consent was provided by parents and children prior to enrollment, and participants were compensated $30 for their time. Data acquisition MR imaging data were acquired at the UC Davis Imaging Research Center with a Siemens 3 T Siemens Trio scanner using a 32-channel head coil. The location of the hippocampus was identified bilaterally using a sagittally acquired localizing scan. A T2-weighted image of the hippocampal formation was acquired perpendicularly to the long axis of the hippocampus using a spin-echo sequence (interleaved acquisition; matrix: 161 mm × 200 mm; in-plane resolution: 0.4 mm × 0.4 mm; slices: 28; slice thickness: 1.9 mm; TR: 4200 ms; TE: 106 ms). This T2 imaging protocol was previously used with adults in the same scanner (Libby et al., 2012). One T1-weighted image was acquired sagittally using a magnetization-prepared rapid acquisition gradient echo (MPRAGE) pulse sequence (matrix: 256 mm × 256 mm; in-plane resolution: 1.0 mm × 1.0 mm; slices: 208; slice thickness: 1.0 mm; TR: 1900 ms; TE: 2.9 ms). Segmentation of hippocampal subfields Independent raters manually segmented all 39 participants' hippocampi bilaterally, resulting in separate volumes for the CA1, a joint region including the CA3 and DG (CA3/DG), and the subiculum. Segmentation was based on a protocol described in Ekstrom et al. (2009) and Zeineh et al. (2001), integrated with guidelines included in Yushkevich et al. (2010), Duvernoy (2005), and Insausti (2010). We used the ITK-SNAP image viewer and segmentation tool (www.itksnap.org) to view and segment the images. To ensure that raters traced subfields under similar

163

viewing conditions, raters adjusted contrast levels in ITK-SNAP so that low intensity white matter voxels were seen as black and high intensity CSF voxels were seen as white. Each rater was blind to age, gender, and memory performance of participants. Segmentation of subfields was completed for each slice within the hippocampal body. The subfields in the head and tail regions were not segmented due to concerns that partial-volume artifacts, which can occur in images with a 1.9 mm slice thickness, would prevent reliable segmentation. The boundary between the head and the body was identified based on the presence of the uncal apex: the body section began one slice posterior to the uncal apex. The body was further segmented from the tail of the hippocampus one slice anterior to the coronal slice at which the fornix separates from the hippocampus in the tail (Watson et al., 1992). Following identification of the hippocampal body, segmentation of subfields continued caudally from the first to the last slice, using the following guidelines (Fig. 1). Segmentation of each coronal slice began with the subiculum, then the CA1, and ended with the CA3/DG. The inferior boundary of the subiculum from the parahippocampal cortex was demarcated at the nadir of the concavity in the medial wall between the collateral sulcus and hippocampus (segment A), which lies approximately midway between the collateral sulcus and the hippocampus. The boundary between the subiculum and CA1 was demarcated by a segment perpendicular to the gray matter ribbon at the point where the hippocampus pinches downward to form a teardrop shape (segment B), which also corresponds to the medial extent of the CA3/DG region. The CA3/DG boundary with the CA1 was delineated by using the following procedure adapted from Yushkevich et al. (2010). First, the longest line from the most medial and inferior extent of the CA3/DG to the most lateral point of the CA fields was drawn (segment C). At the midpoint of segment C, a perpendicular line was drawn superiorly that terminates at the most superior extent of the CA field (segment D). From the superior end of segment D, a perpendicular segment was drawn laterally (segment E) for a length approximately equal to the thickness of the local CA field. Finally, from the lateral end of segment E, we completed the ‘hang-man’ shape by dropping a perpendicular segment to the inferior extent of the local CA field (segment F). Segment F marks the boundary between the CA1 and CA3/DG. The alveus and fimbria were excluded from hippocampal segmentations. The internal stratified laminae identified as the visible lowintensity voxels inside the hippocampus marked the transition from the CA1 and CA3/DG. These low-intensity voxels were included within the CA1 segmentation. The overall hippocampal body volume was also computed by summing the volumes from the CA3/DG, CA1, and subiculum segmentations. We acknowledge that there are additional subfields (e.g., CA2, CA4; Duvernoy, 2005); however, most protocols including our own do not address these subfields. Reliability of segmentation method Before examining associations between subfields, age, and episodic memory, we took several steps to establish inter-rater reliability between two independent tracers. First, intraclass correlation coefficients (ICC; Bartko, 1966) were computed using a two-way random effects model for consistency of averaged measures. Additionally, inter-rater Dice Similarity Coefficients (DSCs; Dice, 1945) were computed to assess the absolute agreement between tracers in terms of volume and spatial position. DSC for each subfield and each participant is computed as follows: DSC = 2|A∩B| / (|A| + |B|), where A and B are the segmentation volumes provided by each of the independent tracers A and B respectively, and |A∩B| is the volume shared by both A and B. Thus, the DSC measures the proportion of spatial overlap between two raters. The coefficient ranges from 0 to 1, where 0 indicates no agreement between tracers, and 1 indicates perfect agreement in both volume and spatial position. It has been argued that DSC ≥ 0.7 represents good to excellent agreement (Bartko, 1991; Zijdenbos et al., 1994). In developmental

164

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

Fig. 1. A. Depiction of demarcation rules in three example slices drawn from the anterior third (top), middle third (middle), and posterior third (bottom) of the hippocampal body. The inferior border of the subiculum is marked at segment A. The subiculum and CA1 are separated at segment B where the hippocampus pinches to form a tear-drop shape. The longest possible segment C is drawn from the superior endpoint of segment B to the most lateral point of the CA field. Segment D is drawn from the midpoint of segment C to the superior extent of the CA field. Segment E is drawn laterally and perpendicularly to segment D; its length is equal to the thickness of the CA field. Finally, segment F is drawn inferiorly and parallel to segment D; its length is also equal to the thickness of the CA field. Segment F marks the boundary between the CA1 and CA3/DG. The low-intensity voxels inside the hippocampus marked the transition from CA1 and CA3/DG, but were included to the CA1 subfield. B. Color rendering of each subfield following demarcation: CA1 appears in yellow, CA3/DG appears in blue, and subiculum appears in purple. The example slices were drawn from the scan of an 8-year-old female participant.

studies, it is important to establish that reliability does not differ as a function of age. Therefore, we tested for age-related differences in DSC. Adjusting the volumes of hippocampal subfields for subsequent analyses The volumes of each subfield as traced by raters were adjusted by estimated intracranial volume (ICV). ICV estimates were derived from T1-weighted structural images of each participant using the FMRIB Software Library v 4.1 (www.fmrib.ox.ac.uk/fsl; Zhang et al., 2001). The skulls were stripped using the Brain Extraction Tool (BET) in a twostep process with an intermediary bias correction. Once the brains were extracted and visually verified for abnormality, each skull-stripped brain was linearly aligned to the MNI152 template. Then, the inverse of the determinant of the affine matrix was obtained and multiplied by the size of the MNI152 template, which gives us the estimated intracranial brain volume. All subfield volumes were then corrected by ICV using the analysis of covariance approach described in Raz et al. (2005), providing independently adjusted volumes for each subfield. Episodic memory assessment Participants performed a task assessing the ability to retain associations between items and arbitrarily colored borders (Fig. 2). Participants viewed 75 line drawings presented sequentially on a laptop; drawings were taken from the age-normed database (Cycowicz et al., 1997), and were 300 × 300 pixels. Each drawing had a 50 pixel red or green colored border. Drawings and their colored borders appeared for 3000 ms, and were followed by a 3000 ms inter-trial fixation cross (Fig. 2A).

Participants were asked to try to remember the drawing and the color of the border that went with each drawing. To do so, they were instructed to think about the identity of each line drawing and to indicate the color of the border with which each line drawing appeared by pressing a corresponding red or green button on the keyboard. Prior to actual encoding, children were familiarized with the procedure in a practice phase with a set of practice line drawings not used during the task. Memory was tested during functional scanning (Fig. 2B). Functional results are not presented in this report. The memory test probes included 75 studied and 75 new drawings. Participants were instructed to determine whether they had seen the drawing with a red border (by pressing the ‘red’ button), with a green border (by pressing the ‘green’ button), or whether the item had not been studied at all (by pressing the ‘new’ button). As part of the functional imaging task, participants intermittently performed an active baseline task appearing pseudorandomly between test trials. In this task, children saw a series of Xs, each appearing at one of three locations on the screen. Participants were to press the button corresponding to each location as each X appeared on the screen (Stark and Squire, 2001). Prior to entering the MR scanner, children were instructed on the rules of the memory retrieval task and the X-game. Participants' understanding of the retrieval task was probed by the experimenter. If deemed necessary, instructions were repeated. Following instruction, participants practiced the retrieval task in a practice phase. The practice phase was identical to the retrieval phase that would occur in the scanner but used line drawings from the practice encoding phase as practice old items and completely novel items as practice distractors. The practice phase allowed participants to

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

165

Fig. 2. Depiction of the item–color episodic memory task. A. At study, participants encode arbitrary relations between a line drawing and the color of a border (red or green). B. At test, participants are asked to remember the color of the border associated with a line drawing, or reject line drawings as unstudied.

acclimate to the retrieval procedure and to map responses to the appropriate buttons. Immediately following practice retrieval, children were placed in the scanner. Versions of this task have been used in previous research; performance on these tasks has been shown to relate to typically developing hippocampal structure (DeMaster et al., 2013) and function (DeMaster and Ghetti, 2012; Ghetti et al., 2010a), and has been found to be sensitive to subtle purported hippocampal damage (Ghetti et al., 2010b). Agerelated improvements in item-recognition and associative memory are typically found across middle childhood and adolescence. We computed hit-rates to studied items, false-alarm rates to unstudied items, and computed the non-parametric measure of recognition memory discrimination A′ (Pollack and Norman, 1964). Our primary measure of associative memory is the proportion given by the frequency of correct retrieval of the item–color associations divided by the sum of correctly identified studied items (aka item–color accuracy). One additional approach, suggested during the review process, is that a measure of associative memory that incorporates the false alarm rate. Therefore, we additionally computed item–color A′ discrimination scores (aka item–color A′). Finally, we assessed IQ with the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999) to ensure that predicted associations between subfields and episodic memory were not due to more general variations in cognitive functioning.

Results Inter-rater reliability The first index of reliability was obtained with ICC analysis across all participants. Results suggest excellent reliability in the CA3/DG and CA1 segmentations in both the left and right hippocampal body. Subiculum segmentations achieved moderate reliability, which is consistent with several previous reports (Mueller et al., 2007; Yushkevich et al., 2010): Left CA1 ICC =0.95, Right CA1 ICC = 0.95; Left CA3/DG ICC = 0.93, Right CA3/DG ICC = 0.95; and Left subiculum ICC = 0.71, Right subiculum ICC =0.81. The overall hippocampal body was computed by summing CA3/DG, CA1, and subiculum segmentations, and maintained excellent reliability coefficients: Left ICC = 0.95 and Right ICC = 0.96. The Fisher–Bonett test was used to test for significant differences in ICC coefficients between subfields within and across the left and right hippocampus (Bonett, 2003; Kim and Feldt, 2008); significance was set at p b .05, and was left uncorrected for multiple comparisons to ensure that we could detect reliability differences should there be any. The subiculum was significantly less reliable than CA3/DG (Left, Z = 3.06, p = .001; Right, Z = 2.87, p = .002), and CA1 (Left, Z = 3.78, p =.000; Right, Z = 2.87, p = .002) segmentations. CA3/DG and CA1 coefficients did not differ reliably (Left, Z = 0.72, p = .235; Right, Z = 0.04, p = .483).

166

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

We then examined DSCs for the left and right hippocampal segmentations to assess the degree of spatial agreement between tracers. DSC coefficients between subfields and hippocampi were tested using an independent samples t-test, with an uncorrected pvalue b .05 (Fig. 3). High levels of spatial agreement were achieved. Left CA1 DSC = 0.85, Right CA1 DSC = 0.86; Left CA3/DG DSC = 0.84, Right CA3/DG DSC =0.85; Left subiculum DSC = 0.79, Right subiculum DSC = 0.77. Overall, our findings for DSCs followed the pattern of results seen with the ICC analyses. DSC coefficients were significantly smaller in the left subiculum in comparison to the left CA1 t(38) = 8.56, p b .05, and the left CA3/DG t(38) =5.67, p b .05, as well as in the right subiculum in comparison to the right CA1 t(38) = 12.50, p b .05 and the right CA3/DG t(38) = 10.88, p b .05. DSC coefficients for the overall hippocampal body were excellent: Left Hippocampus DSC = 0.90 and Right Hippocampus DSC = 0.90. DSC coefficients did not differ between the left and right hippocampal segmentations. Overall, these results indicate high levels of inter-rater agreement. To establish whether similar levels of reliability were evident in younger compared to older children, we divided the sample into quartiles based on age (Quartile 1: 8–8.95 years, n = 10; Quartile 2: 8.96–11 years, n = 10; Quartile 3: 11.01–13.53 years, n = 10; Quartile 4: 13.52–14.9 years, n = 9), and conducted a 4 (age group: Quartile 1 vs. Quartile 2 vs Quartile 3 vs Quartile 4) × 3 (subfield: CA1 vs CA3/DG vs subiculum) × 2 (hemisphere: right versus left) mixed analysis of variance (ANOVA). Results confirmed across all ages the effect of field, F (2, 70) = 105.22, p b .001, η 2p = .75, such that DSCs for the subiculum were significantly lower than those for the CA1 and CA3/DG bilaterally. No significant effect of age group, F (3, 35) = .71, p = .55, η2p = .06, or hemisphere, F (1, 35) = .06, p = .81, η2p = .002, was found (Fig. 3). Thus, our results showed that there were no differences in inter-rater agreement as a function of age.

Developmental differences in subfield volumes In all analyses, subfield volumes were first adjusted for intracranial volume as described in the Materials and methods section. To test for age differences in subfield volumes, we divided the sample into quartiles based on age as we did in the reliability analyses (Quartile 1: 8–8.95 years, n = 10; Quartile 2: 8.96 to 11 years, n = 10; Quartile 3: 11.01–13.53 years, n = 10; Quartile 4: 13.52 to 14.9 years, n = 9). Means and standard deviations for the left and right subfields of each age group are listed in Table 1. We conducted a series of 4 (age quartiles) one-way ANOVAs, with each of the subfields as the dependent measure. We further tested linear and quadratic polynomial contrasts in the hippocampal subfields. Results are shown in Fig. 4. On the right hippocampus, significant age differences were found in the CA3/DG, F (3, 35) = 3.18, p = .036, η2p = .214, and CA1, F (3, 35) = 3.32, p = .031, η2p = .221, but not in the subiculum, F (3, 35) = .70, p = .560, η2p = .056. As shown in Fig. 4, age-related increases are observed in the first three age groups, but not in the oldest group. Indeed children in the third quartile (11 to 13 years of age) exhibit larger volumes than all of the other child groups, ps b .05. These differences were captured by significant quadratic contrast in the right CA3/DG, F (1, 35) = 4.47, p = .042, and the right CA1, F (1, 35) = 8.49, p = .006, but not in the right subiculum, F (1, 35) = 1.79, p = .189. Linear contrasts were not statically significant in these subfields, F (1, 35) ≤ .324, p ≥ .773. In contrast, no significant age differences were found in any of the fields in the left hippocampus, Fs (3, 35) ≤ 2.11, ps ≥ .122, η2ps ≤ .150. However, the overall pattern of results in part resembled those found on the right hippocampus. Indeed, similar to the right hippocampus, a significant quadratic contrast was observed in left CA1, F (1, 35) = 5.07, p = .031, but not in the left subiculum, F (1, 35) = 0.24, p = .630. Unlike the right CA3/DG, the left CA3/DG did not show a significant age-related quadratic contrast, F (1, 35) = 2.28, p = .140. No age-related linear

Fig. 3. Inter-rater reliability as measured by Dice coefficients as a function of age quartile (Quartile 1, 8–8.95 years, n = 10; Quartile 2, 8.96 to 11 years, n = 10; Quartile 3, 11.01– 13.53 years, n = 10; Quartile 4, 13.52 to 14.9 years, n = 9) and subfield in the left (A) and right hippocampus (B). Error bars correspond to 95% confidence intervals.

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

167

Table 1 Volumes (mm3) of subfields in the left and right hippocampal body across age quartiles: Quartile 1, 8–8.95 years; Quartile 2, 8.96–11 years; Quartile 3, 11.01–13.53; Quartile 4, 13.52–14. 9 years. Hemisphere

Subfield

Left hippocampus

CA1 CA3/DG Subiculum Body CA1 CA3/DG Subiculum Body

Right hippocampus

1st quartile (n = 10)

2nd quartile (n = 10)

3rd quartile (n = 10)

4th quartile (n = 9)

Mean

SD

Mean

SD

Mean

SD

Mean

SD

941 667 727 2336 954 690 713 2358

122 145 118 289 158 180 116 419

1116 691 770 2578 1086 703 759 2548

273 219 89 496 187 154 118 332

1145 780 812 2737 1182 866 790 2839

219 185 142 473 237 210 195 550

1032 640 821 2494 962 649 720 2332

133 74 82 190 148 96 75 231

contrast reached statistical significance, although the left subiculum closely approached it, F (1, 35) ≤ 4.08, p = .051. As a final analysis, we examined age-related differences in the overall volume of the hippocampal body. On the right, there was not a significant age-related linear contrast, F (1, 35) = 0.23, p = .633, but there was a significant age-related quadratic contrast, F (1, 35) = 7.13, p = .011. The overall left hippocampal body did not reveal a significant linear contrast, F (1, 35) = 1.45, p = .237, or quadratic contrast F (1, 35) = 3.77, p = .060. Associations between subfield volumes and episodic memory The memory task used in the present study resulted in above chance performance levels (hit rate, M = 70, SD = .17; false-alarm rate, M = .21, SD = .22; A′, M = .82, SD = .12, item–color association rate, M = .59, SD = .11) even though children in Quartile 1 exhibited significantly lower hit rates (M = .59, SD = .17) compared to children in Quartile 2 (M = .77, SD = .13, p b .05) and Quartile 4 (M = .75, SD = .12, p b .05), but not children in Quartile 3 (M = .71, SD = .21, p = .10). No other significant differences were found, ps ≥ .12. We note that four participants did not contribute data to this analysis because they elected not to complete the behavioral task. Given the age-related differences in subfields reported above, associations between episodic memory measures and subfields were conducted using partial correlations which controlled for the effect of age. Significant relationships between the subfield volume of the right hippocampus and memory performance were observed. Specifically, the right CA3/DG positively correlated with item–color accuracy, r(32) = .35, p = .046, and item–color A′, r(32) = .42, p = .015. The right CA3/DG volume negatively correlated with false alarm rates, r(32) = − .44, p = .009; see Fig. 5. While the partial correlations

between the right subiculum and item–color accuracy did not reach significance, r(32) = .21, p = .245, the right subiculum positively correlated with item–color A′, r(32) = .36, p = .036, and negatively correlated with false alarm rate r(32) = −.37, p = .029. The partial correlations between the right CA1 and memory performance were not significant (item–color accuracy, r(32) = .20, p = .268, item–color A′, r(32) = .18, p = .316, and false-alarms r(32) = − .25, p = .159). The partial correlation between the overall right hippocampal body (sum of the CA1, CA3/DG and subiculum) and item–color accuracy did not reach conventional levels of statistical significance, r(32) = .30, p = .090. Item–color A′ measure, which incorporated the false alarm rate, retained its statistical significance correlated with the overall volume in the right hippocampal body, r(32) = .36, p = .035, as did the rate of false alarms, r(32) = − .40, p = .016. Neither volumes of the left hippocampal subfields nor of the overall left hippocampal body significantly correlated with item–color accuracy, |r(32)| ≤ .32, p ≥ .066, item–color A′, |r(32)| ≤ .17, p ≥ .32, or with false alarm rates |r(32)| ≤ .25, p ≥ .159. No subfield in either hemisphere significantly correlated with item hit rate, |r(32)| ≤ .30, p ≥ .088, or with A′ measure of the overall recognition memory discrimination, |r(32)| ≤ .31, p ≥ .077. Correlations retained their statistical significance when we controlled for the effect of IQ. Discussion To our knowledge, no published study has examined age-related differences in hippocampal subfields during childhood or the relation between subfield volumes and episodic memory during this period. In the present research, we first validated a high-resolution imaging method and protocol for hippocampal subfield segmentation in a pediatric sample. Second, we sought to investigate age-related differences in

Fig. 4. Age-related differences in subfield volume as a function of age-quartile (Quartile 1, 8–8.95 years, n = 10; Quartile 2, 8.96–1 years, n = 10; Quartile 3, 11.01–13.53 years, n = 10; Quartile 4, 13.52–14. 9 years, n = 9) and subfield in the left and right hippocampus. Error bars correspond to standard errors.

168

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

hippocampal functioning by showing correlations between subfield volumes and memory (Engvig et al., 2012; Mueller et al., 2011; Shing et al., 2011), and by identifying structural changes due to hippocampal disturbance (Engvig et al., 2012; Mueller, and Weiner, 2009; Shing et al., 2011). However, up until now high-resolution imaging methods have not been used or validated with pediatric populations. We followed procedures described in Ekstrom et al. (2009) and Zeineh et al. (2001), with contributions by Yushkevich et al. (2010), and with guidance from hippocampal atlases by Duvernoy (2005) and Insausti (2010). Using these methods, we found overall excellent levels of reliability across subfields and hemispheres comparable to those reported in research with adults using similar imaging and tracing protocols to our own (Mueller et al., 2007; Yushkevich et al., 2010). Furthermore, the absence of age-related differences in reliability provides direct evidence that the current protocol can be successfully used with children as young as 8. Future research should examine whether high-resolution imaging of younger children would pose new and different challenges. We acknowledge that manual segmentations only included the hippocampal body. This choice was due to limitations in imaging resolution and contrast in the hippocampal head and tail. However, this approach is common in the literature (e.g., Chen et al., 2011; Das et al., 2012; Mueller et al., 2011; Olsen et al., 2013; Palombo et al., 2013) and in fact our study includes more portions of the body than several other studies in adults have included (Mueller and Weiner, 2009; Shing et al., 2011). Overall, our results suggest that this protocol could be used to address substantive questions about subfield development and relation with episodic memory in childhood. Development of hippocampal subfields

Fig. 5. Plots of partial correlations between memory for memory performance and the volume of the right CA3/DG, controlling for age. A. Correlation between the right CA3/ DG volume and rate of item–color memory. Z-scores for each measure are plotted. B. A significant negative correlation between the CA3/DG volume and false alarms. Z-scores for each measure are plotted.

the volume of hippocampal subfields, given recent evidence that structural development of the hippocampus occurs into adolescence (Abrahám et al., 2010; Gogtay et al., 2006; Insausti, 2010). Third, we sought to investigate associations between volumes of the subfields and episodic memory in childhood while controlling for the effect of age to test the hypothesis that the CA3/DG would be associated with episodic memory requiring item–context associative binding (Kesner, 2007; Kirwan and Stark, 2007; Rolls, 2010). The results addressing these goals are discussed in turn. Segmentation of hippocampal subfields A sizable literature exists on human hippocampal subfields including a number of studies using high-resolution imaging methods (e.g., Ekstrom et al., 2009; Mueller et al., 2007; Wisse et al., 2012; Zeineh et al., 2001). These studies have contributed to our understanding of

Recent research has begun to highlight that the hippocampus might exhibit protracted development extending into adolescence, and that this development is heterogeneous in nature with different trajectories observed along the longitudinal axis of the hippocampus (Abrahám et al., 2010; DeMaster et al., 2013; Gogtay et al., 2006; Insausti, 2010). The cytoarchitecture of the hippocampus may be another important dimension of change. Subfields are not equally distributed along the longitudinal axis (e.g., Duvernoy, 2005), and may in part account for the developmental differences reported previously (e.g., DeMaster et al., 2013). Consistent with this idea, previous post-mortem research has shown that the hippocampal body increases in volume with age, and that the DG and the CA fields show more robust and uniform volumetric increases than all subregions of the subicular complex (Insausti, 2010). In the present study, we showed that both the CA3/DG and CA1 increased in volume from 8 to 13 years of age. We further note that these data are largely consistent with reports of a protracted trajectory in DG and CA3 volumes in macaque primates (Jabès et al., 2010). The source of the apparent decrease in volume after age 13 is unclear, but is consistent with curvilinear age/volume relations reported for hippocampal subregions along the anterior/posterior axis (Gogtay et al., 2006). In contrast, no reliable differences were found in the subiculum. This finding is consistent with prior post-mortem findings (Insausti, 2010), and provides initial in vivo evidence that hippocampal subfields might follow distinct developmental trajectories, and that the quadratic trajectory observed by Gogtay and colleagues may be attributable to the CA1 and CA3/DG, but not to the subiculum. Alternatively, the lack of age-related differences observed in the subiculum may be attributed to the lower inter-rater reliability achieved in the left and the right subiculum. Significant age differences in the subfields were primarily found in the right hippocampus, except for the significant quadratic contrast in the left CA1. Though there are no strong a priori reasons to predict hemispheric differences in hippocampal development, these results are consistent with other work indicating more robust developmental differences in the right compared to the left hippocampus (DeMaster et al., 2013; Gogtay et al., 2006).

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

One could speculate that the volumetric differences in the CA3/DG may be a result of processes of neurogenesis in the dentate gyrus. Increased and decreased rates of neurogenesis have been associated with increased or decreased volumetric changes in the hippocampus, respectively (e.g., Sahay et al., 2007; Snyder et al., 2009), which might be associated with changes in memory function (e.g., Nakashiba et al., 2012). Our current imaging method, however, does not allow us to separate the DG from the CA3, thus preventing us from connecting our results specifically to changes in the DG. As imaging protocols improve it will be possible to test this hypothesis more directly. Associations between subfields and episodic memory We had predicted that the DG and CA3 would be associated with episodic memory and indeed found a positive relationship between the volume of the combined CA3/DG regions in the right hippocampal body and successful retrieval of the association between an item and the color of its border. One hypothesized function of the DG is to disambiguate representations sharing similar features and characteristics thereby enhancing specificity of encoded memories (McNaughton and Morris, 1987). This would result in increased likelihood of subsequent accurate retrieval and decreased likelihood of inaccurate retrieval (Hunsaker and Kesner, 2013; Kirwan and Stark, 2007; Rolls, 2010; Shing et al., 2011). Further, the CA3 is thought to support encoding and retrieval of event features and their relations (e.g., Hunsaker and Kesner, 2013; Kesner, 2007; Rolls, 2010). The positive relation between the CA3/DG volume and memory is consistent with either of these hypotheses about DG or CA3 function. Moreover, we found a negative partial correlation between the CA3/ DG volume and the rate of false alarms to new line drawings while controlling for age and intracranial volume. This study was not designed with the necessary power to examine potential interactions between age and the relationship between the CA3/DG volume and episodic memory performance. Future studies are planned to address this limitation. The present result is consistent with the DG encoding distinct memory traces resistant to interference or over-generalization in children and adolescents, as has been reported in adult populations (e.g., Shing et al., 2011; Stark et al., 2013). It is also possible that this partial correlation reflects novelty detection or retrieval functions in this region (e.g., Bakker et al., 2008; Duncan et al., 2012; Knight, 1996; Nakashiba et al., 2012). Future studies should differentiate these possibilities during development. We could not distinguish between DG and CA3 fields in the present study, and both were expected to be related positively to episodic memory; however, as high-resolution methods improve, it will be important to distinguish the contribution of these two subfields to the development of episodic memory. Future research will need to examine how the dentate gyrus and CA fields uniquely contribute to episodic memory. In contrast, CA1 volume was not significantly associated with episodic memory. While it is clear that the CA1 subfield contributes to episodic memory (e.g., Zola-Morgan et al., 1986; Bartsch et al., 2011), one hypothesized function within the hippocampal circuitry could relate to processing temporal attributes of episodes (e.g., event sequences; Hunsaker and Kesner, 2013; Hunsaker et al., 2008) and their relation with other attributes of the context (Eichenbaum, 2013; MacDonald et al., 2011). Given that the temporal sequence of studied items was irrelevant to task success, the task may not have been sensitive to the functional contribution of the CA1 to episodic memory. However, we cannot exclude that statistical power may have reduced our ability to uncover a significant relation with memory. Finally, the volume of the right subiculum was negatively associated with false alarms, positively associated with item–color A′ discrimination scores, but failed to reach conventional levels of statistical significance in predicting memory for item–color accuracy, item recognition hit rate or A-prime measure of item-recognition discrimination. The

169

primary difference between item–color accuracy measure and the item–color A′ score is the incorporation of the false alarm rate into the score calculation. Thus, the false-alarm rate is likely responsible for the differing result between item–color accuracy and item–color A′ and subiculum volume. Relatively little is known about the function of the subiculum. One possibility is that the subiculum might support processes of novelty detection during encoding (Preston et al., 2010) and retrieval (Bakker et al., 2008; Zeineh et al., 2003). The current results are consistent with a role for the subiculum in novelty detection, at least as it pertains to the capacity to detect and reject novel items. The position of the subiculum in the hippocampal circuit suggests that it may play a role in outputting signals to the rest of the brain. For example, the results of a recent rodent study (Kim et al., 2012) suggest that the subiculum compresses sparse hippocampal representations to a distributed and information-rich representation appropriate for communication with other brain regions. If this is the case, volumetric measures of the subiculum should correlate with overt memory measures in a way that largely overlaps with the way in which other volumes correlate. This was partly the case in the current study, where the right subiculum volume was negatively correlated with false-alarm rates.

Conclusions The present research is the first to establish the feasibility of highresolution methods with children and adolescents and to provide a first examination of age-related differences in hippocampal subfield volumes during this developmental period. Further, this research is the first to relate subfield volumes to episodic memory in childhood or adolescence. These associations were observed despite relatively limited age differences in memory ability in the present study. Agerelated differences in the subfields suggest that the developmental contribution of the hippocampus to episodic memory extends into adolescence. Examining these differences offers a chance to better understand the contribution of function of subfields to episodic memory and its development. We acknowledge that episodic memory emerges from the contribution of a developing network of brain regions (Ofen, 2012; Ghetti & Bunge, 2012). For example, the protracted development of the prefrontal cortex and the mnemonic control processes it supports may account for many of the developmental improvements in episodic memory. The present study does not attempt to characterize the contribution of the hippocampus to episodic memory while accounting for the contributions made by other regions in the medial temporal lobe, prefrontal cortex and parietal lobes, or their interactions. We agree that network level analyses are needed in the future to situate any observed age-related changes in the hippocampus within the larger network of brain regions thought to support episodic memory. In most likelihood, regions within the network develop symbiotically. These are challenging questions, and the present study achieves its primary goals: (1) to demonstrate reliability of high resolution methods in children, (2) to examine trajectories of development in the volume of hippocampal subfields in children and adolescents, and (3) to examine the correspondence between subfield volume and episodic memory, apart from the effect of age.

Acknowledgments This research was supported by a Scholar Award from the James S. McDonnell Foundation to S.G. We thank Abbie Thompson for her assistance with subfield tracing. Conflict of interest The authors declare no competing financial interests.

170

J.K. Lee et al. / NeuroImage 94 (2014) 162–171

References Abrahám, H., Vincze, A., Jewgenow, I., Veszprémi, B., Kravják, A., Gömöri, E., Seress, L., 2010. Myelination in the human hippocampal formation from midgestation to adulthood. Int. J. Dev. Neurosci. 28, 401–410. Bakker, A., Kirwan, C.B., Miller, M., Stark, C.E., 2008. Pattern separation in the human hippocampal CA3 and dentate gyrus. Science 319, 1640–1642. Bartko, J.J., 1966. The intraclass correlation coefficient as a measure of reliability. Psychol. Rep. 19, 3–11. Bartko, J.J., 1991. Measurement and reliability: statistical thinking considerations. Schizophr. Bull. 17, 483–489. Bartsch, T., Döhring, J., Rohr, A., Jansen, O., Deuschl, G., 2011. CA1 neurons in the human hippocampus are critical for autobiographical memory, mental time travel, and autonoetic consciousness. Proc. Natl. Acad. Sci. 108, 17562–17567. Bonett, D.G., 2003. Sample size requirements for comparing two alpha coefficients. Appl. Psychol. Meas. 27, 72–74. Chen, J., Olsen, R.K., Preston, A.R., Glover, G.H., Wagner, A.D., 2011. Associative retrieval processes in the human medial temporal lobe: hippocampal retrieval success and CA1 mismatch detection. Learn. Mem. 18, 523–528. Cycowicz, Y.M., Friedman, D., Rothstein, M., Snodgrass, J.G., 1997. Picture naming by young children: norms for name agreement, familiarity, and visual complexity. J. Exp. Child Psychol. 65, 171–237. Das, S.R., Avants, B.B., Pluta, J., Wang, H., Suh, J.W., Weiner, M.W., Mueller, S.G., Yushkevich, P.A., 2012. Measuring longitudinal change in the hippocampal formation from in vivo high-resolution T2-weighted MRI. Neuroimage 60, 1266–1279. DeMaster, D.M., Ghetti, S., 2012. Developmental differences in hippocampal and cortical contributions to episodic retrieval. Cortex 49, 1482–1493. DeMaster, D., Pathman, T., Lee, J.K., Ghetti, S., 2013. Structural development of the hippocampus and episodic memory: developmental differences along the anterior/ posterior axis. Cereb. Cortex. http://dx.doi.org/10.1093/cercor/bht160. Dice, L.R., 1945. Measures of the amount of ecologic association between species. Ecology 26, 297–302. Duncan, K., Ketz, N., Inati, S.J., Davachi, L., 2012. Evidence for area CA1 as a match/mismatch detector: a high‐resolution fMRI study of the human hippocampus. Hippocampus 22, 389–398. Duvernoy, H.M., 2005. The Human Hippocampus: Functional Anatomy, Vascularization and Serial Sections with MRI. Springer. Eichenbaum, H., 2013. Memory on time trends. Cogn. Sci. 17, 81–88. Eichenbaum, H., Cohen, N.J., 2001. From Conditioning to Conscious Recollection: Memory Systems of the Brain. Oxford University Press. Ekstrom, A.D., Bazih, A.J., Suthana, N.A., Al-Hakim, R., Ogura, K., Zeineh, M., Burggren, A.C., Bookheimer, S.Y., 2009. Advances in high-resolution imaging and computational unfolding of the human hippocampus. Neuroimage 47, 42–49. Engvig, A., Fjell, A.M., Westlye, L.T., Skaane, N.V., Sundseth, Ø., Walhovd, K.B., 2012. Hippocampal subfield volumes correlate with memory training benefit in subjective memory impairment. Neuroimage 61, 188–194. Ghetti, S., Lee, J.K., 2011. Children's episodic memory. Wiley Interdiscip. Rev. Cogn. Sci. 2, 365–373. Ghetti, S., DeMaster, D.M., Yonelinas, A.P., Bunge, S.A., 2010a. Developmental differences in medial temporal lobe function during memory encoding. J. Neurosci. 30, 9548–9556. Ghetti, S., Lee, J.K., Sims, C.E., Demaster, D.M., Glaser, N.S., 2010b. Diabetic ketoacidosis and memory dysfunction in children with type 1 diabetes. J. Pediatr. 156, 109–114. Giedd, J.N., Vaituzis, A.C., Hamburger, S.D., Lange, N., Rajapakse, J.C., Kaysen, D., Vauss, Y.C., Rapoport, J.L., 1996. Quantitative MRI of the temporal lobe, amygdala, and hippocampus in normal human development: ages 4–18 years. J. Comp. Neurol. 366, 223–230. Gogtay, N., Nugent, T.F., Herman, D.H., Ordonez, A., Greenstein, D., Hayashi, K.M., Clasen, L., Toga, A.W., Giedd, J.N., Rapoport, J.L., Thompson, P.M., 2006. Dynamic mapping of normal human hippocampal development. Hippocampus 16, 664–672. Hunsaker, M.R., Kesner, R.P., 2013. The operation of pattern separation and pattern completion processes associated with different attributes or domains of memory. Neurosci. Biobehav. Rev. 37, 36–58. Hunsaker, M.R., Lee, B., Kesner, R.P., 2008. Evaluating the temporal context of episodic memory: the role of CA3 and CA1. Behav. Brain Res. 188, 310–315. Insausti, R., 2010. Postnatal development of the human hippocampal formation. Adv. Anat. Embryol. Cell 206. Insausti, R., Amaral, D.G., 2004. Hippocampal formation. Hum. Nerv. Syst. 871–914. Jabès, A., Lavenex, P.B., Amaral, D.G., Lavenex, P., 2010. Quantitative analysis of postnatal neurogenesis and neuron number in the macaque monkey dentate gyrus. Eur. J. Neurosci. 31, 273–285. Kalkan, Y., Unal, B., Keles, O.N., Kara, A., 2013. Numerical analysis of age and genderdependent neuronal cells in postnatal development of rat hippocampus. Neurol. Psychiatry Brain Res. 19, 19–28. Kesner, R.P., 2007. Behavioral functions of the CA3 subregion of the hippocampus. Learn. Mem. 14, 771–781. Kim, S., Feldt, L.S., 2008. A comparison of tests for equality of two or more independent alpha coefficients. J. Educ. Meas. 45, 179–193. Kim, S.M., Ganguli, S., Frank, L.M., 2012. Spatial information outflow from the hippocampal circuit: distributed spatial coding and phase precession in the subiculum. J. Neurosci. 32, 11539–11558. Kirwan, C.B., Stark, C.E.L., 2007. Overcoming interference: an fMRI investigation of pattern separation in the medial temporal lobe. Learn. Mem. 14, 625–633. Knight, R.T., 1996. Contribution of human hippocampal region to novelty detection. Nature 383, 256–259.

Lavenex, P.B., Lavenex, P., 2013. Building hippocampal circuits to learn and remember: insights into the development of human memory. Behav. Brain Res. 254, 8–21. Leutgeb, J.K., Leutgeb, S., Moser, M.B., Moser, E.I., 2007. Pattern separation in the dentate gyrus and CA3 of the hippocampus. Science 315, 961–966. Libby, L.A., Ekstrom, A.D., Ragland, J.D., Ranganath, C., 2012. Differential connectivity of perirhinal and parahippocampal cortices within human hippocampal subregions revealed by high-resolution functional imaging. J. Neurosci. 32, 6550–6560. MacDonald, C.J., Lepage, K.Q., Eden, U.T., Eichenbaum, H., 2011. Hippocampal “time cells” bridge the gap in memory for discontiguous events. Neuron 71, 737–749. McNaughton, B.L., Morris, R.G., 1987. Hippocampal synaptic enhancement and information storage within a distributed memory system. Trends Neurosci. 10, 408–415. Mueller, S.G., Weiner, M.W., 2009. Selective effect of age, Apo e4, and Alzheimer's disease on hippocampal subfields. Hippocampus 19, 558–564. Mueller, S.G., Stables, L., Du, A.T., Schuff, N., Truran, D., Cashdollar, N., Weiner, M.W., 2007. Measurement of hippocampal subfields and age-related changes with high resolution MRI at 4 T. Neurobiol. Aging 28, 719–726. Mueller, S.G., Chao, L.L., Berman, B., Weiner, M.W., 2011. Evidence for functional specialization of hippocampal subfields detected by MR subfield volumetry on high resolution images at 4 T. Neuroimage 56, 851–857. Nakashiba, T., Cushman, J.D., Pelkey, K.A., Renaudineau, S., Buhl, D.L., McHugh, T.J., Barrera, V.R., Chittajallu, R., Iwamoto, K.S., McBain, C.J., Fanselow, M.S., Tonegawa, S., 2012. Young dentate granule cells mediate pattern separation, whereas old granule cells facilitate pattern completion. Cell 149, 188–201. Ofen, N., 2012. The development of neural correlates for memory formation. Neurosci. Biobehav. Rev. 36, 1708–1717. Ofen, N., Shing, Y.L., 2013. From perception to memory: changes in memory systems across the lifespan. Neurosci. Biobehav. Rev. 37, 2258–2267. Olsen, R.K., Palombo, D.J., Rabin, J.S., Levine, B., Ryan, J.D., Rosenbaum, R.S., 2013. Volumetric analysis of medial temporal lobe subregions in developmental amnesia using high-resolution magnetic resonance imaging. Hippocampus 23, 855–860. Østby, Y., Tamnes, C.K., Fjell, A.M., Westlye, L.T., Due-Tønnessen, P., Walhovd, K.B., 2009. Heterogeneity in subcortical brain development: a structural magnetic resonance imaging study of brain maturation from 8 to 30 years. J. Neurosci. Off. J. Soc. Neurosci. 29, 11772–11782. Palombo, D.J., Amaral, R.S.C., Olsen, R.K., Müller, D.J., Todd, R.M., Anderson, A.K., Levine, B., 2013. KIBRA polymorphism is associated with individual differences in hippocampal subregions: evidence from anatomical segmentation using high-resolution MRI. J. Neurosci. 33, 13088–13093. Pollack, L., Norman, D.A., 1964. Non-parametric analysis of recognition experiments. Psychon. Sci. 1, 125–126. Preston, A.R., Bornstein, A.M., Hutchinson, J.B., Gaare, M.E., Glover, G.H., Wagner, A.D., 2010. High-resolution fMRI of content-sensitive subsequent memory responses in human medial temporal lobe. J. Cogn. Neurosci. 22, 156–173. Raz, N., Lindenberger, U., Rodrigue, K.M., Kennedy, K.M., Head, D., Williamson, A., Dahle, C., Gerstorf, D., Acker, J.D., 2005. Regional brain changes in aging healthy adults: general trends, individual differences and modifiers. Cereb. Cortex 15, 1676–1689. Rolls, E.T., 2010. A computational theory of episodic memory formation in the hippocampus. Behav. Brain Res. 215, 180–196. Sahay, A., Drew, M.R., Hen, R., 2007. Dentate gyrus neurogenesis and depression. Prog. Brain Res. 163, 697–722. Sahay, A., Scobie, K.N., Hill, A.S., O'Carroll, C.M., Kheirbek, M.A., Burghardt, N.S., Fenton, A. A., Dranovsky, A., Hen, R., 2011. Increasing adult hippocampal neurogenesis is sufficient to improve pattern separation. Nature 472, 466–470. Shing, Y.L., Rodrigue, K.M., Kennedy, K.M., Fandakova, Y., Bodammer, N., Werkle-Bergner, M., Lindenberger, U., Raz, N., 2011. Hippocampal subfield volumes: age, vascular risk, and correlation with associative memory. Front. Aging Neurosci. 3, 2. Snyder, J.S., Radik, R., Wojtowicz, J.M., Cameron, H.A., 2009. Anatomical gradients of adult neurogenesis and activity: young neurons in the ventral dentate gyrus are activated by water maze training. Hippocampus 19, 360–370. Stark, C.E.L., Squire, L.R., 2001. When zero is not zero: the problem of ambiguous baseline conditions in fMRI. Proc. Natl. Acad. Sci. 98, 12760–12766. Stark, S.M., Yassa, M.A., Lacy, J.W., Stark, C.E., 2013. A task to assess behavioral pattern separation (BPS) in humans: data from healthy aging and mild cognitive impairment. Neuropsychologia 51, 2442–2449. Watson, C., Andermann, F., Gloor, P., Jones-Gotman, M., Peters, T., Evans, A., Olivier, A., Melanson, D., Leroux, G., 1992. Anatomic basis of amygdaloid and hippocampal volume measurement by magnetic resonance imaging. Neurology 42, 1743–1750. Wechsler, D., 1999. Wechsler Abbreviated Scale of Intelligence. Psychological Corporation. Wisse, L.E.M., Gerritsen, L., Zwanenburg, J.J.M., Kuijf, H.J., Luijten, P.R., Biessels, G.J., Geerlings, M.I., 2012. Subfields of the hippocampal formation at 7 T MRI: in vivo volumetric assessment. Neuroimage 61, 1043–1049. Yassa, M.A., Lacy, J.W., Stark, S.M., Albert, M.S., Gallagher, M., Stark, C.E.L., 2011. Pattern separation deficits associated with increased hippocampal CA3 and dentate gyrus activity in non-demented older adults. Hippocampus 21, 968–979. Yu, E.P., Dengler, C.G., Frausto, S.F., Putt, M.E., Yue, C., Takano, H., Coulter, D.A., 2013. Protracted postnatal development of sparse specific dentate granule cell activation in the mouse hippocampus. J. Neurosci. 33, 2947–2960. Yurgelun-Todd, D.A., Killgore, W.D.S., Cintron, C.B., 2003. Cognitive correlates of medial temporal lobe development across adolescence: a magnetic resonance imaging study. Percept. Mot. Skills 96, 3–17. Yushkevich, P.A., Wang, H., Pluta, J., Das, S.R., Craige, C., Avants, B.B., Weiner, M.W., Mueller, S., 2010. Nearly automatic segmentation of hippocampal subfields in in vivo focal T2-weighted MRI. Neuroimage 53, 1208–1224. Zeineh, M.M., Engel, S.A., Thompson, P.M., Bookheimer, S.Y., 2001. Unfolding the human hippocampus with high resolution structural and functional MRI. Anat. Rec. 265, 111–120.

J.K. Lee et al. / NeuroImage 94 (2014) 162–171 Zeineh, M.M., Engel, S.A., Thompson, P.M., Bookheimer, S.Y., 2003. Dynamics of the hippocampus during encoding and retrieval of face–name pairs. Science 299, 577–580. Zhang, Y., Brady, M., Smith, S., 2001. Segmentation of brain MR images through a hidden Markov random field model and the expectation–maximization algorithm. IEEE Trans. Med. Imaging 20, 45–57. Zijdenbos, A.P., Dawant, B.M., Margolin, R.A., Palmer, A.C., 1994. Morphometric analysis of white matter lesions in MR images: method and validation. IEEE Trans. Med. Imaging 13, 716–724.

171

Zola-Morgan, S., Squire, L.R., Amaral, D.G., 1986. Human amnesia and the medial temporal region: enduring memory impairment following a bilateral lesion limited to field CA1 of the hippocampus. J. Neurosci. 6, 2950–2967.