Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Oct 17.
Published in final edited form as: Proc SPIE Int Soc Opt Eng. 2015 Dec 22;9681:96810W. doi: 10.1117/12.2211383

Low-frequency fluctuation amplitude analysis of resting-state fMRI in sickle cell disease

Julie Coloigner a,b, Yeun Kim a,b, Adam Bush d, Matt Borzage b,c,d, Vidya Rajagopalan a,b,e, Natasha Lepore a,b, John Wood d
PMCID: PMC6192018  NIHMSID: NIHMS977539  PMID: 30344362

Abstract

Sickle cell disease may result in neurological damage and strokes, leading to morbidity and mortality. Currently, there are no dependable biomarkers to predict impending strokes. In this study, we analyzed neuronal processes at resting state and more particularly how this disease affects the default mode network. The amplitude of low frequency fluctuations was used to reflect areas of spontaneous BOLD signal across brain regions. We compared the activations of sickle cell disease patients to a control group using variance analysis and t-test. Significant regional differences among the two groups were observed, especially in the default mode network areas and cortical regions near large cerebral arteries. These findings suggest that sickle cell disease causes activation modifications near vessels, and these changes could be used as a biomarker of the disease.

Keywords: Functional magnetic resonance imaging, Resting state, Connectivity, Sickle cell disease, Amplitude of low-frequency fluctuations

1. INTRODUCTION

Sickle cell disease (SCD) is a chronic blood disorder caused by a mutation in the hemoglobin beta gene that affects around 50 million of people throughout the world, particularly descendants of Sub-Saharan Africans. Red blood cells lack the flexibility required to flow in the circulatory system, leading to vasoocclusion, ischemia and infarcts. More particularly, SCD can cause neurological damage and strokes, leading to significant mortality and morbidity. Identification of cerebral biomarkers could be useful to predict brain alterations and to diagnose disease states.1 However, structural magnetic resonance imaging is not sensitive enough to detect preclinical biomarkers that may aid in SCD disease stratification. This has highlighted the need to investigate other neuroimaging markers in the brain of SCD patients.2–4

Functional Magnetic Resonance Imaging (fMRI) is a non-invasive technique for measuring and mapping the brain activity. Spontaneous low frequency fluctuations of Blood Oxygenation Level-Dependent (BOLD) signal have been observed during rest and can be identified with baseline spontaneous activity in the brain. The default mode network (DMN) which consists of the precuneus, posterior cingulate cortex, medial prefrontal cortex and lateral parietal cortex5 is of particular interest. Indeed, in a growing number of studies, the activity within the DMN is analyzed to understand how spontaneous synchronized activity of neuronal populations is altered in diseases such as depression, Alzheimers diseases, schizophrenia or epilepsy.6,7

In this paper, we analyze the resting state functional magnetic resonance signal, whose low-frequency fluctuations are considered to reflect spontaneous synchronized activity of neuronal populations. The Amplitude of Low Frequency Fluctuations (ALFF)8 is used to determine areas of spontaneous BOLD signal across brain regions. Given the cognitive deficits and abnormal blood flow in SCD patients, we hypothesize that significant alterations in DMN activation patterns exist between patients with SCD and control subjects. Furthermore, we postulate that abnormal ALFF patterns in SCD may predict some of the neurocognitive deficits found in these patients.

2. MATERIAL AND METHODS

2.1 Dataset

This study was approved by institutional review board and conducted in the Children’s Hospital Los Angeles (CHLA) between January 2012 and May 2015. Written informed consent was obtained from all subjects. Since 40% of SCD patients develop stroke by 20 years of age, we restricted our population to adolescents and young adults without strokes. 20 SCD patients (age=23 ± 8) and 17 control subjects (age=21 ± 1) were included in this analysis. Exclusion criteria included pregnancy, previous stroke, acute chest or paincrisis hospitalization within one month. The patient groups were matched in terms of sex and education. All participants underwent a battery of tests: structural and functional imaging and neurophysiological tests.

MRI data were acquired on a 3T Philips Achieva at CHLA. A 3D T1-weighted image was acquired covering the whole brain (160 sagittal slices) with TR = 8.20 s, TE = 3.77 ms, flip angle = 8, in-plane resolution = 256×256, FOV = 256 mm × 224 mm and thickness/gap =1.0/0 mm). During resting-state fMRI scanning, subjects were instructed to close their eyes, keep still as much as possible, not to think of anything systematically and not to fall asleep. The functional images were acquired with the following parameters: TR = 2000 ms, TE = 50 ms, flip angle = 90, in-plane resolution = 96 × 96, FOV = 220 mm × 220 mm, 26 axial slices, thickness/gap= 5/0 mm. A total of the 240 volumes were collected in 8 minutes.

2.2 Preprocessing steps

The resting-state fMRI data were preprocessed with the FMRIB Software Library (FSL), using standard spatial preprocessing methods. The first 5 volumes from each subject were discarded to allow the signal to reach equilibrium and to allow the participant to adapt to the scanning noise. The remaining 235 volumes were corrected for the acquisition time delay between slices. Rigid realignment was then performed to remove physiological motion such as high-frequency respiratory and cardiac noise, with FSLs MCFLIRT. A motion index, DVARS, which represents the change in BOLD signal from volume to volume, was then computed. This measure was calculated as the root mean squared value of the differentiated time series (by backwards differences). Some volumes were excluded because of excessive DVARS.

Then, a two-step coregistration method was used to transform the fMRI data into the Montreal Neurological Institute (MNI) template space. First, the mean fMRI image was coregistered (6 parameters) with the 3D T1-weighted image; then the resulting image was linearly transformed to the Montreal Neurological Institute (MNI) template space, using the linear and non-linear algorithms of FSL, called FLIRT and FNIRT, respectively. The generated parameters for these two coregistration steps were then concatenated and used for the normalization of each fMRI volume. Finally, the registered fMRI volumes were normalized and smoothed with a Gaussian kernel of 8mm × 8 mm × 8mm full-width at half maximum (FWHM).

After fMRI preprocessing, in order to remove residual motion, the 6 linear parameters estimated by FSL’s MCFLIRT, as well as their derivatives (6 parameters) were used as regressors, computed by backwards differences (12 regressors in total). Moreover, the physiological noise was also reduced using other nuisance regressors which were calculated by the CompCor9 method as implemented in Nilearn (http://nilearn.github.io/). These 5 regressors correspond to the principal components from noisy regions-of-non-interest, such as white matter, cerebral spinal fluid, and matter outside the brain. Then, the fMRI data were temporally band-pass filtered (0.009 < f < 0.08 Hz) to reduce low-frequency drift and higher frequency respiratory and cardiac noise.

2.3 ALFF analysis

The ALFF analysis10 was performed using FSL functions. The power spectrum was obtained after the time series for each voxel were transformed to the frequency domain with a Fast Fourier Transform (FFT). Since the power of a given frequency is proportional to the square of the amplitude of this frequency component of the original time series in the time domain, the square root was calculated at each frequency of the power spectrum. The averaged square root was obtained across 0.009 − 0.08 Hz for each voxel. This averaged square root was taken as the ALFF. To reduce global effects of variability across the participants, the ALFF value at each voxel location was divided by the global mean ALFF value. The individual map was transformed to a Z score (i.e., we subtracted the global mean value, and then divided by the standard deviation) instead of being simply divided by the global mean.

2.4 Statistical Analysis

2.4.1 Within-group ALFF analysis

To explore the within-group whole-brain ALFF patterns, one-sample t-tests were performed on the individual ALFF maps in a voxelwise manner for the two groups: SCD and control. A gray matter mask of the MNI template was used. The corrected threshold was determined using the AlphaSim program11 with a p-value threshold set at P < 0.05 and a cluster size > 1.91 cm3.

2.4.2 Between-group ALFF analysis

To investigate the variability of the activation, one-way analysis of variance (ANOVA) was performed to determine the ALFF differences among the two groups at each voxel (within a gray matter mask). The statistical map was corrected again using the AlphaSim program, with the threshold set at P < 0.05 and a cluster size larger than 1.91 cm3. After identifying all significant clusters, we made a mask including these regions. A post hoc t-test was then performed to compare the Z-maps of the activations between the patient and control groups, restricting the analysis within this mask.

3. RESULTS

As illustrated in the Figure 1, within each group, significant higher ALFF was found in several brain regions, such as medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC), precuneus and posterior frontal cortex.

Figure 1.

Figure 1

Open in a new tab

Representative one-sample t-test results of ALFF maps (P < 0.05, AlphaSim corrected) in groups of SCD (A) and control (B) subjects. Color scale indicates t values. Thresholds were set at a corrected p < 0.05, determined by Monte Carlo simulation. The planes are X = 0, Y = 50 and Z = 35 for the sagittal, coronal and axial views, respectively, in the Montreal Neurological Institute (MNI) template.

The one-way ANOVA revealed 8 clusters with significant differences across the SCD and control groups, including medial orbitofrontal cortex (MOFC), anterior cingulate cortex (ACC), Brodmann Area 8 (BA 8), posterior region of middle cingulate cortex, posterior cingulate cortex (PCC), anterior lobe of cerebellum, posterior lobe of cerebellum and parietooccipital sulcus, as shown in table 1 and figure 2. Table 1 illustrates the results of the t-test for these 8 regions. Compared with the control subjects, the SCD patients showed a significant decrease in ALFF in PCC, posterior region of cingulate cortex and MOFC. In addition, an increased activity in ACC, BA 8, cerebellum and parietooccipital sulcus was found in the sickle cell disease group.

Table 1.

List of brain areas with significant differences in ALFF during the resting state between the SCD and control groups. A positive sign in the peak t-value represents an increase, while a negative one represents a decrease in SCD group. A corrected threshold of p < 0.05 determined by Monte Carlo simulation was taken as meaning that there was a significant difference between groups for either t-test and ANOVA test.

Cerebral region MNI coordinates Cluster volume (mm3) F-value t-value
(1) Decreased regions
Medial orbitofrontal cortex [−4,30, −12] 28950 11.94 −3.76
Posterior region of middle cingulate cortex [−2,26,36] 1542 6.81 −1.82
Posterior cingulate cortex [−2, −23,34] 1467 6.72 −2.64
(2) Increased regions
Anterior cingulate cortex [−2,30,15] 1542 5.11 3.09
Brodmann Area 8 [2,19,43] 7.03 6.46 2.95
Anterior lobe of cerebellum [−3, −60, −12] 1011 9.58 2.99
Posterior lobe of cerebellum [−6, −78, −27] 1317 8.30 1.90
Parietooccipital sulcus [20, −74,35] 1304 5.53 3.19
Open in a new tab

Figure 2.

Figure 2

Open in a new tab

Representative one-way ANOVA results of ALFF maps (P < 0.05, AlphaSim corrected) between SCD and control groups. On the sagittal view (higher, left), six clusters are indicated by arrows (clockwise): medial orbitofrontal cortex (1), anterior cingulate cortex (2), Brodmann Area 8 (BA 8) (3), posterior region of cingulate cortex (4), posterior cingulate cortex (5) and anterior lobe of cerebellum (6). In the axial views, the regions indicated by arrows are as follows: posterior cingulate cortex (Z=34), posterior region of middle cingulate cortex (Z=33), anterior cingulate cortex (Z=6), medial orbitofrontal cortex (Z=−9). In the coronal view, the anterior lobe of cerebellum (6) and parietooccipital sulcus (7) are designated by an arrow.

4. DISCUSSION

Using the ALFF metric of resting state fMRI, several neuronal activity changes were detected in specific brain regions in SCD patients versus control subjects. These neuronal abnormalities were found in 8 areas in the SCD group. This may be due to a variety of factors, including for example fear extinction, tissue damage from strokes, and vasculature differences that are not apparent on MR imaging.

The analysis of ALFF indicated that patients with sickle cell disease showed differences in activity within regions of the DMN. First, we observed that resting-state functional connectivity of the ACC was increased in patients with SCD when compared to control subjects. Most SCD patients suffer from episodic or chronic pain, which may influence cortical ALFF. Indeed, previous studies have found a strong association between ACC and the emotional response to pain.12 Buchel et. al.13 found a positive correlation between fMRI signal intensity and pain intensity. SCD patients experience recurrent, severe pain crises that may partly contribute to the abnormal increased activity in ACC. We are performing pain assessment questionaires in our patients and will compare those results with differences in ACC activation.

Second, we observed a decreased activity in the posterior components of the DMN, the PCC and precuneus, which are thought to act as intrinsic mediatory nodes of this network.14 Regional differences in cortical thickness from SCD patients were also located on the posterior medial surfaces, the so-called ”watershed” areas between the perfusion territories of the anterior cerebral artery (ACA) and the posterior cerebral artery (PCA).15 The precuneus and the posterior cingulate are vulnerable to acute anemia and, thus, may be susceptible to abnormal development.

The orbitofrontal cortex covers the ventral surface of the primate prefrontal cortex. Several animal studies as well as human neuroimaging studies have reported that the medial orbitofrontal cortex (mOFC) and the ventromedial prefrontal cortex (vmPFC) are associated with fear extinction.16–21 We found these areas have lower ALFF values in the SCD patients than in the control group. We attribute this phenomenon to the heightened level of fear and anxiety of the control group due to their unfamiliarity to hospital exams and environment. Since the SCD patients are more used to hospital setting and exams, the lower ALFF value may reflect these differences.

Third, we observed four regions with abnormal activity in the SCD group: Brodmann Area 8, anterior lobe of cerebellum, posterior lobe of cerebellum and parietooccipital sulcus. In fMRI studies, BOLD signals are often affected by vascular activities, causing the signals to fluctuate. Thus, the ALFF values of these four regions may be affected by vasculature, since they are located near large cerebral arteries such as the ACA, the superior cerebellar artery, the anterior inferior cerebellar artery and the parieto-occipital branch of posterior cerebral artery, respectively.

5. CONCLUSION

In this paper, we studied the brain functional activation, using the ALFF analysis. Our comparison among the SCD and control groups confirmed our hypothesis about the difference in resting state activity for SCD patients, and revealed a difference in DMN and cortical regions near large cerebral arteries. Our results suggest that the effects of the cerebrovascular disease associated with SCD produces specific changes in brain activity. We could potentially use activity intensities of certain brain regions, such as precuneus, to measure the amount of pre-existing hypoxic brain damage, thereby allowing us to evaluate the severity of the disease. The next step will consist in taking into account different measures in the study, such as the neurocognitive scores and physiological parameters. Moreover, the activity value of ACC could potentially serve as an objective clinical measure of pain, after correlating them to pain assessment scales.

References

  • 1.Elias E, Liao J, Jara H, Watanabe M, Nadgir R, Sakai Y, Erbay K, Saito N, Ozonoff A, Steinberg M, Sakai O. Quantitative mri analysis of craniofacial bone marrow in patients with sickle cell disease. J Neuroradiol. 2013;34(3):622–7. doi: 10.3174/ajnr.A3240. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Steen RG, Emudianugh T, Hunte M, Glass J, Wu S, Xiong X, Reddick W. Brain volume in pediatric patients with sickle cell disease: evidence of volumic growth delay? AJNR Am J Neuroradiol. 2005;26(3):455–62. [PMC free article] [PubMed] [Google Scholar]
  • 3.Kirk GR, Haynes MR, Palasis S, Brown C, Burns TG, McCormick M, Jones RA. Regionally specific cortical thinning in children with sickle cell disease. Cerebral Cortex. 2009;19:1549–1556. doi: 10.1093/cercor/bhn193. [DOI] [PubMed] [Google Scholar]
  • 4.Balci A, Karazincir S, Beyoglu Y, Cingiz C, Davran R, Gali E, Okuyucu E, Egilmez E. Quantitative brain diffusion-tensor mri findings in patients with sickle cell disease. American Journal of Roentgenology. 2012;198(5):1167–74. doi: 10.2214/AJR.11.7404. [DOI] [PubMed] [Google Scholar]
  • 5.Raichle M, MacLeod A, Snyder A, Powers W, Gusnard D, Shulman G. A default mode of brain function. Proc Natl Acad Sci. 2001;98(2) doi: 10.1073/pnas.98.2.676. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Fox M, Greicius M. Clinical applications of resting state functional connectivity. Front Syst Neuroscience. 2010:4–19. doi: 10.3389/fnsys.2010.00019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Qi Z, Wu X, Wang Z, Zhang N, Dong H, Yao L, Li K. Impairment and compensation coexist in amnestic mci default mode network. Neuroimage. 2010;50:48–55. doi: 10.1016/j.neuroimage.2009.12.025. [DOI] [PubMed] [Google Scholar]
  • 8.Zou QH, Zhu CZ, Yang Y, Zuo XN, Long XY, Cao QJ, Wang YF, Zang YF. An improved approach to detection of amplitude of low-frequency fluctuation (alff) for resting state fmri: Fractional alff. J of Neuroscience Methods. 2008;172:137–141. doi: 10.1016/j.jneumeth.2008.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Behzadi Y, Restom K, Liau J, Liu T. A component based noise correction method (compcor) for bold and perfusion based fmri. Neuroimage. 2007;375(1):90–101. doi: 10.1016/j.neuroimage.2007.04.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Qi R, Zhang L, Wu S, Zhong J, Ni L, Zhang Z, Li K, Jia Q, Wu X, Liu Y, Lu G. Altered resting-state brain activity at functional mr imaging during the progression of hepatic encephalopathy. Radiology. 2012;264 doi: 10.1148/radiol.12111429. [DOI] [PubMed] [Google Scholar]
  • 11.Ledberg A, Akerman S, Roland P. Estimation of the probabilities of 3D clusters in functional brain images. Neuroimage. 1998;8:113–128. doi: 10.1006/nimg.1998.0336. [DOI] [PubMed] [Google Scholar]
  • 12.Baliki MN, Mansour AR, Baria AT, Apkarian AV. Functional reorganization of the default mode network across chronic pain conditions. Plos One. 2014 doi: 10.1371/journal.pone.0106133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Buchel C, Bornhvd K, Quante M, Glauche V, Bromm B, Weiller C. Dissociable neural responses related to pain intensity, stimulus intensity, and stimulus awareness within the anterior cingulate cortex: a parametric single-trial laser functional magnetic resonance imaging study. The Journal of Neuroscience. 2002;22:970–976. doi: 10.1523/JNEUROSCI.22-03-00970.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Cavanna A, Trimble M. The precuneus: a review of its functional anatomy and behavioural correlates. Brain. 2006;129:564–83. doi: 10.1093/brain/awl004. [DOI] [PubMed] [Google Scholar]
  • 15.Kirk G, Haynes M, Palasis S, Brown C, Burns T, McCormick M, Jones R. Regionally specific cortical thinning in children with sickle cell disease. Cereb Cortex. 2009;19:1549–56. doi: 10.1093/cercor/bhn193. [DOI] [PubMed] [Google Scholar]
  • 16.Milad M, Rauch S, Pitman R, Quirk G. Fear extinction in rats: implications for human brain imaging and anxiety disorders. Biol Psychol. 2006;73:61–71. doi: 10.1016/j.biopsycho.2006.01.008. [DOI] [PubMed] [Google Scholar]
  • 17.Barad M. Fear extinction in rodents: basic insight to clinical promise. Curr Opin Neurobiol. 2005;15:710–715. doi: 10.1016/j.conb.2005.10.005. [DOI] [PubMed] [Google Scholar]
  • 18.Sotres-Bayon F, Cain C, Ledoux J. Brain mechanisms of fear extinction: historical perspectives on the contribution of prefrontal cortex. Biol Psychiatry. 2006;60:329–336. doi: 10.1016/j.biopsych.2005.10.012. [DOI] [PubMed] [Google Scholar]
  • 19.Rauch S, Shin L, Phelps E. Neurocircuitry models of posttraumatic stress disorder and extinction: Human neuroimaging research past, present and future. Biol Psychiatry. 2006;60:376–382. doi: 10.1016/j.biopsych.2006.06.004. [DOI] [PubMed] [Google Scholar]
  • 20.Quirk G, Garcia R, Gonzalez-Lima F. Prefrontal mechanisms in extinction of conditioned fear. Biol Psychiatry. 2006;60:337–343. doi: 10.1016/j.biopsych.2006.03.010. [DOI] [PubMed] [Google Scholar]
  • 21.Davis M, Myers K, Chhatwal J, Ressler K. Pharmacological treatments that facilitate extinction of fear: relevance to psychotherapy. NeuroRx. 2006;3:82–96. doi: 10.1016/j.nurx.2005.12.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES