- Email: [email protected]
P-104
Posters: P-101 P-101
A THREE-YEAR FOLLOW-UP SURVEY OF PATIENTS WITH MILD COGNITIVE IMPAIRMENT (MCI): AJIMU PROJECT
Mika Sugimura, Seigo Nakano, Tatsuo Yamada, Fukuoka University, School of Medicine, Fukuoka, Japan. Contact e-mail: [email protected] Background: It is significant to conduct a longitudinal study on the cognitive functions of MCI patients in a community. Objective(s): To conduct a three-year follow-up survey of 64 senior citizens aged 65 or over in Ajimu district who were diagnosed with amnestic MCI (MCI with memory loss only) from June 2003 to November 2004. Methods: 1251 elderly citizens aged 65 and over in Ajimu district were assessed for their cognitive functions using a method called “Five Cog,” a neuropsychological assessment comprising the following five categories: memory, attention, verbal fluency, clock drawing test, and analogy. 64 seniors, who had been found to have memory deficits, were selected for the subjects of this study. After 3 years of follow-up, 27 who could visit Ajimu health center underwent the Five Cog examination and were reevaluated by a doctor. 37 who could not visit the center were reevaluated in the doctor’s individual visits and underwent MMSE examination. Results: Of the 64 subjects, 26 were males (age, mean⫾SD; 74.6⫾6.8, education period⫾SD; 10.3⫾2.0 years), and 38 were females (age, mean 74.7⫾7.5, education period, 10.0⫾2.1). After 3 years of follow-up, 21 maintained the state of MCI and 22 were diagnosed as normal. 18 were diagnosed with dementia (Alzheimer’s disease: 16, Dementia with Lewy bodies: 1, Mixed dementia: 1). 3 were dead, 2 of whom were considered to have developed Alzheimer’s disease before their death by the interviews with their families. A nonpharmacological intervention was given to 18 subjects to prevent dementia. In this group, 16 shifted to normal, 2 maintained the state of MCI and no one shifted to dementia. Conclusions: The nonpharmacological intervention appears effective to prevent MCI patients from developing dementia. P-102
COGNITIVE EFFECTS OF MEMANTINE IN MODERATE TO SEVERE ALZHEIMER’S DISEASE: A RESPONDER ANALYSIS
Michael Tocco1, E. Malca Resnick1, Stephen M. Graham1, Vojislav Pejovic2, 1Forest Research Institute, Jersey City, NJ, USA; 2 Prescott Medical Communications Group, Chicago, IL, USA. Contact e-mail: [email protected] Background: Memantine is a moderate-affinity, uncompetitive N-methyl-Daspartate (NMDA) receptor antagonist approved in the US and Europe for the treatment of moderate to severe Alzheimer’s disease (AD). In a previously published, 28-week, double-blind, placebo-controlled trial in moderate to severe AD, memantine-treated patients (n⫽126) showed significantly better cognitive performance on the Severe Impairment Battery (SIB) than placebotreated patients (n⫽126). Methods: In this post hoc analysis, treatment groups were compared in terms of the percentage of patients who showed a cognitive response on the SIB total score and three SIB subscales (memory, language, and praxis) across the trial and at endpoint. We examined the number of patients who demonstrated any response to treatment (improvement or no decline compared to baseline; SIB score change ⱖ0), a strong response (score change ⱖ4), and a very strong response (score change ⱖ8). Analysis was performed on the observed cases of the intention-to-treat population using a generalized estimating equations model. Results: On the SIB total score, the proportion of patients who demonstrated any response in the memantine group was significantly higher compared to the placebo group across the entire trial (P⫽.005), and at study endpoint (37.5% vs. 22.9%; P⫽.048). Similarly, the proportion of strong responders was significantly higher in the memantine group across the entire trial (P⬍.001), and at study endpoint (19.8% vs. 6.0%; P⫽.015), while the proportion of very strong responders was significantly higher in the memantine group across the entire trial (P⫽.014), and supported memantine treatment at study endpoint (10.4% vs. 2.4%; P⫽.06). On the SIB subscales, a higher proportion of responders was observed in the memantine group across the entire trial for Memory (strong response: P⫽.04), Language
S131
(strong response: P⫽.031), and Praxis (any response: P⫽.010; strong response: P⫽.033). At study endpoint, significant memantine response was demonstrated for Language (any response: 38.5% vs. 24.1%; P⫽.027) and Praxis (any response: 50.5% vs. 30.1%; P⫽.004) subscales. Conclusions: In patients with moderate to severe AD, a significantly greater percentage of memantine-treated patients demonstrated improvement or no cognitive decline relative to untreated patients. Similar results were seen when individual cognitive domains of memory, language, and praxis were examined. P-103
THERAPY FOLLOW UP IN MILD DEMENTIA SUBJECTS
Veronika Vachapov1, Ruth Verchovsky1, Tali Vishne2, Amos D. Korczyn3, 1Sourasky Medical Center, Tel Aviv, Israel; 2Bar ilan University, Givat Shmuel, Israel; 3Tel Aviv University, Tel Aviv, Israel. Contact e-mail: [email protected] Background: Although there is no cure for Alzheimer’s disease, some medications seem to delay the progress of the disease. To test the effect of these medications, cognitive assessments are used, most of which are limited either by low sensitivity (i.e. MMSE) or by high complexity and cost (i.e. neuropsychologist’s evaluation). Computerized neurocognitive tests are an affordable and efficient solution for treatment monitoring, however, most batteries are not useful for patients with no prior computer skills. Objective(s): We used a novel self-administrated computerized test for monitoring the therapy process of incipient or mild dementia subjects. Methods: Twelve elderly subjects, 64-87 years old were recruited from the memory disorders clinic in the Sourasky medical center. All had good or corrected vision and hearing and could understand Hebrew, English or Russian. MMSE scores’ range was 21-27. Seven subjects were computer naive. After baseline cognitive test, using NexAde, a self-administered computerized neuropsychological battery, subjects received Rivastigmine tartrate (n⫽6) or Donepezil HCl (n⫽6) for 21 - 189 days. Analysis used general linear model with repeated measure with treatment duration, age, sex, MMSE, and medication type as covariates. Results: Age (F⫽.457, p⫽.833), gender (F⫽.457, p⫽.833), and computer skill (F⫽2.250, p⫽.208) of the subjects did not influenced their performance in the seven different NexAde subtests. Performance in different subtests was correlated to MMSE scores at baseline (F⫽15.323, p⫽.017). MMSE results did not change with treatment (F⫽.017, p⫽.902). On the other hand, test results improved with treatment: the longer was the treatment, the better were the test results (F⫽14.885, p⫽.018). This difference was more prominent in visual identification, visual recall, and digit span backward tests. There was no difference between the performance of patients taking different medications (F⫽.801, p⫽510). Conclusions: Our study confirmed previous results on modest cognitive improvement in mild to moderate dementia patients. While the computerized battery used in the current study was sensitive enough to measure the improvement, MMSE failed to do so. Different cognitive functions were influenced differently by the treatment implying high variance in treatment response. Further study is needed to identify subjects who will or will not respond to treatment beforehand. P-104
REPRODUCIBILITY OF CORTICAL THICKNESS ASSESSMENT ACROSS SCANNER UPGRADE IN THE AGES: REYKJAVIK STUDY
Alex P. Zijdenbos1,2, Jason P. Lerch3,1, Sigurdur Sigurdsson4, Alan C. Evans2,1, Vilmundur Gudnason4, Lenore J. Launer5, 1Neuralyse Inc., Montreal, PQ, Canada; 2McGill University, Montreal, PQ, Canada; 3SickKids, Toronto, ON, Canada; 4Icelandic Heart Association, Kopavogur, Iceland; 5National Institute on Aging, Bethesda, MD, USA. Contact e-mail: [email protected] Background: Quantitative measures of cortical thickness (CT) are showing promise for the assessment of disease progression in dementia, but these methods need to be evaluated in light of changes in MRI hardware and software that routinely occur. The Age, Gene/Environment Susceptibility Reykjavik Study (AGES-RS) is a population-based study (n⫽5764) of ageing.
S132
Posters: P-105
Brain MRI is acquired on a dedicated 1.5T GE Signa Twinspeed MR scanner. One quarter of the way through the study the system was upgraded to a GE Signa Twinspeed EXCITE, which included a change in a software platform from v9.0 to v10.0 and a change in a RF array head coil from a 4-channel- to an 8-channel cap coil. To establish data uniformity across this upgrade, a cohort of 32 AGES-RS subjects was scanned twice before the upgrade, and twice after (n⫽128 scans). Objective(s): To demonstrate that CT is reproducible across the scanner upgrade performed in AGES-RS. Methods: CT was measured independently in the 128 scans using the methods described in [1-3]. Statistical analyses were performed both for average cortical thickness as well as for each vertex of the surface models. Statistical significance was assessed using a 5% false discovery rate. Results: There was no change in mean cortical thickness (p⫽0.84) or the variance between scans (p⫽0.81) as a result of the scanner upgrade. When tested at every vertex of the surface models, the only difference in CT was found in the occipital lobe. Conclusions: Reproducibility of computer algorithms is essential to establish in studies of dementia. The measurement of cortical thickness (CT) as described in [2,3] is robust against variations in scanner configurations. The small regional effect found in the occipital lobe in this study is consistent with the location of MRI signal changes that result from differences in receiver coil geometry. Preliminary experiments, results of which will be presented, indicate that this occipital effect can be eliminated by adapting the parameters of the intensity nonuniformity correction algorithm [4] to account for multi-channel coil designs. This study further establishes MR based cortical thickness algorithms to be reproducible and robust to hardware changes [5,6,7].
[1] Zijdenbos et al., IEEE TMI, 21(2), 2002 [2] Lerch et al., NeuroImage 24, 2005 [3] Ad-Dab’bagh et al., Proc. 12th annual meeting OHBM, 2006 [4] Sled et al., IEEE TMI, 17(1), 1998 [5] Lerch et al., Cerebral Cortex 15(7), 2005 [6] Lee et al., NeuroImage 31, 2006 [7] Han et al., NeuroImage 32, 2006
P-105
THREE-DIMENSIONAL RECONSTRUCTION OF HISTOLOGICAL SECTIONS AND CORRELATIONS WITH MRI FROM TRANSGENIC MODELS OF ALZHEIMER’S DISEASE
M. Mallar Chakravarty1, Barry J. Bedell1,2, Simone P. Zehntner1, Kurt Hemmings1, Edith Hamel1, Alan C. Evans1,2, D. Louis Collins1, 1 Montreal Neurological Institute, Montreal, PQ, Canada; 2Neuralyse Inc., Montreal, PQ, Canada. Contact e-mail: [email protected] Background: Non-invasive imaging modalities, such as MRI and PET, show great promise in longitudinal studies of transgenic mouse models of Alzheimer’s disease (AD). The routing use of imaging markers, however, requires extensive validation against gold-standard, ex vivo tissue studies. To this end, we have developed sophisticated methods to allow for fully-automated quantification of histological measures. Objective: In this abstract, we outline one of the building blocks in our fully-automated quantification process. This includes careful acquisition of serial two-dimensional histological slices (2D), the reconstruction of these serial histological slices into a three-dimensional (3D) histological volume, and the mapping of this volume onto an MRI atlas. Methods: Histological Data Acquisition: All APP mice brains were formalin-fixed and embedded in a specially dyed paraffin block to enhance tissue-to-paraffin contrast. Whole-brain, serial coronal histological sections were prepared and stained on an automated tissue stainer with Luxol fast blue/Hematoxylin & Eosin (H&E), for highcontrast of anatomical structures. The stained sections were digitized using a Zeiss MIRAX Scan ultra-high resolution, high-throughput, whole-slide scanner. Additionally a block-face photograph was acquired for each section and acts as a local reference for the histological data. Reconstruction of Histological Data: The reconstruction of the histological data requires a three-step process. In the first step, the digital blockface images were reconstructed as a contiguous 3D volume. The analogous slice of histological data was then transformed to fit the blockface using a three-parameter linear transformation. In the final step, the slice-to-slice morphological inconsistencies were corrected using 2D nonlinear warping techniques. The histological volume was warped to fit onto an MRI atlas created from the unbiased nonlinear average of ten T2-weighted mouse volumes. Results: Figure 1 depicts the reconstructed histological volume and mapping onto the MRI atlas. Note that the crosshairs identify the same anatomical structures on the in vivo and ex vivo volumes. Conclusions: These novel methods which we have developed will allow for automated, robust correlations of in vivo and ex vivo studies, and provide for rigorous validation of new imaging techniques, as well as the seamless integration of molecular, histopathological and structural/functional imaging studies.
A THREE-YEAR FOLLOW-UP SURVEY OF PATIENTS WITH MILD COGNITIVE IMPAIRMENT (MCI): AJIMU PROJECT
Mika Sugimura, Seigo Nakano, Tatsuo Yamada, Fukuoka University, School of Medicine, Fukuoka, Japan. Contact e-mail: [email protected] Background: It is significant to conduct a longitudinal study on the cognitive functions of MCI patients in a community. Objective(s): To conduct a three-year follow-up survey of 64 senior citizens aged 65 or over in Ajimu district who were diagnosed with amnestic MCI (MCI with memory loss only) from June 2003 to November 2004. Methods: 1251 elderly citizens aged 65 and over in Ajimu district were assessed for their cognitive functions using a method called “Five Cog,” a neuropsychological assessment comprising the following five categories: memory, attention, verbal fluency, clock drawing test, and analogy. 64 seniors, who had been found to have memory deficits, were selected for the subjects of this study. After 3 years of follow-up, 27 who could visit Ajimu health center underwent the Five Cog examination and were reevaluated by a doctor. 37 who could not visit the center were reevaluated in the doctor’s individual visits and underwent MMSE examination. Results: Of the 64 subjects, 26 were males (age, mean⫾SD; 74.6⫾6.8, education period⫾SD; 10.3⫾2.0 years), and 38 were females (age, mean 74.7⫾7.5, education period, 10.0⫾2.1). After 3 years of follow-up, 21 maintained the state of MCI and 22 were diagnosed as normal. 18 were diagnosed with dementia (Alzheimer’s disease: 16, Dementia with Lewy bodies: 1, Mixed dementia: 1). 3 were dead, 2 of whom were considered to have developed Alzheimer’s disease before their death by the interviews with their families. A nonpharmacological intervention was given to 18 subjects to prevent dementia. In this group, 16 shifted to normal, 2 maintained the state of MCI and no one shifted to dementia. Conclusions: The nonpharmacological intervention appears effective to prevent MCI patients from developing dementia. P-102
COGNITIVE EFFECTS OF MEMANTINE IN MODERATE TO SEVERE ALZHEIMER’S DISEASE: A RESPONDER ANALYSIS
Michael Tocco1, E. Malca Resnick1, Stephen M. Graham1, Vojislav Pejovic2, 1Forest Research Institute, Jersey City, NJ, USA; 2 Prescott Medical Communications Group, Chicago, IL, USA. Contact e-mail: [email protected] Background: Memantine is a moderate-affinity, uncompetitive N-methyl-Daspartate (NMDA) receptor antagonist approved in the US and Europe for the treatment of moderate to severe Alzheimer’s disease (AD). In a previously published, 28-week, double-blind, placebo-controlled trial in moderate to severe AD, memantine-treated patients (n⫽126) showed significantly better cognitive performance on the Severe Impairment Battery (SIB) than placebotreated patients (n⫽126). Methods: In this post hoc analysis, treatment groups were compared in terms of the percentage of patients who showed a cognitive response on the SIB total score and three SIB subscales (memory, language, and praxis) across the trial and at endpoint. We examined the number of patients who demonstrated any response to treatment (improvement or no decline compared to baseline; SIB score change ⱖ0), a strong response (score change ⱖ4), and a very strong response (score change ⱖ8). Analysis was performed on the observed cases of the intention-to-treat population using a generalized estimating equations model. Results: On the SIB total score, the proportion of patients who demonstrated any response in the memantine group was significantly higher compared to the placebo group across the entire trial (P⫽.005), and at study endpoint (37.5% vs. 22.9%; P⫽.048). Similarly, the proportion of strong responders was significantly higher in the memantine group across the entire trial (P⬍.001), and at study endpoint (19.8% vs. 6.0%; P⫽.015), while the proportion of very strong responders was significantly higher in the memantine group across the entire trial (P⫽.014), and supported memantine treatment at study endpoint (10.4% vs. 2.4%; P⫽.06). On the SIB subscales, a higher proportion of responders was observed in the memantine group across the entire trial for Memory (strong response: P⫽.04), Language
S131
(strong response: P⫽.031), and Praxis (any response: P⫽.010; strong response: P⫽.033). At study endpoint, significant memantine response was demonstrated for Language (any response: 38.5% vs. 24.1%; P⫽.027) and Praxis (any response: 50.5% vs. 30.1%; P⫽.004) subscales. Conclusions: In patients with moderate to severe AD, a significantly greater percentage of memantine-treated patients demonstrated improvement or no cognitive decline relative to untreated patients. Similar results were seen when individual cognitive domains of memory, language, and praxis were examined. P-103
THERAPY FOLLOW UP IN MILD DEMENTIA SUBJECTS
Veronika Vachapov1, Ruth Verchovsky1, Tali Vishne2, Amos D. Korczyn3, 1Sourasky Medical Center, Tel Aviv, Israel; 2Bar ilan University, Givat Shmuel, Israel; 3Tel Aviv University, Tel Aviv, Israel. Contact e-mail: [email protected] Background: Although there is no cure for Alzheimer’s disease, some medications seem to delay the progress of the disease. To test the effect of these medications, cognitive assessments are used, most of which are limited either by low sensitivity (i.e. MMSE) or by high complexity and cost (i.e. neuropsychologist’s evaluation). Computerized neurocognitive tests are an affordable and efficient solution for treatment monitoring, however, most batteries are not useful for patients with no prior computer skills. Objective(s): We used a novel self-administrated computerized test for monitoring the therapy process of incipient or mild dementia subjects. Methods: Twelve elderly subjects, 64-87 years old were recruited from the memory disorders clinic in the Sourasky medical center. All had good or corrected vision and hearing and could understand Hebrew, English or Russian. MMSE scores’ range was 21-27. Seven subjects were computer naive. After baseline cognitive test, using NexAde, a self-administered computerized neuropsychological battery, subjects received Rivastigmine tartrate (n⫽6) or Donepezil HCl (n⫽6) for 21 - 189 days. Analysis used general linear model with repeated measure with treatment duration, age, sex, MMSE, and medication type as covariates. Results: Age (F⫽.457, p⫽.833), gender (F⫽.457, p⫽.833), and computer skill (F⫽2.250, p⫽.208) of the subjects did not influenced their performance in the seven different NexAde subtests. Performance in different subtests was correlated to MMSE scores at baseline (F⫽15.323, p⫽.017). MMSE results did not change with treatment (F⫽.017, p⫽.902). On the other hand, test results improved with treatment: the longer was the treatment, the better were the test results (F⫽14.885, p⫽.018). This difference was more prominent in visual identification, visual recall, and digit span backward tests. There was no difference between the performance of patients taking different medications (F⫽.801, p⫽510). Conclusions: Our study confirmed previous results on modest cognitive improvement in mild to moderate dementia patients. While the computerized battery used in the current study was sensitive enough to measure the improvement, MMSE failed to do so. Different cognitive functions were influenced differently by the treatment implying high variance in treatment response. Further study is needed to identify subjects who will or will not respond to treatment beforehand. P-104
REPRODUCIBILITY OF CORTICAL THICKNESS ASSESSMENT ACROSS SCANNER UPGRADE IN THE AGES: REYKJAVIK STUDY
Alex P. Zijdenbos1,2, Jason P. Lerch3,1, Sigurdur Sigurdsson4, Alan C. Evans2,1, Vilmundur Gudnason4, Lenore J. Launer5, 1Neuralyse Inc., Montreal, PQ, Canada; 2McGill University, Montreal, PQ, Canada; 3SickKids, Toronto, ON, Canada; 4Icelandic Heart Association, Kopavogur, Iceland; 5National Institute on Aging, Bethesda, MD, USA. Contact e-mail: [email protected] Background: Quantitative measures of cortical thickness (CT) are showing promise for the assessment of disease progression in dementia, but these methods need to be evaluated in light of changes in MRI hardware and software that routinely occur. The Age, Gene/Environment Susceptibility Reykjavik Study (AGES-RS) is a population-based study (n⫽5764) of ageing.
S132
Posters: P-105
Brain MRI is acquired on a dedicated 1.5T GE Signa Twinspeed MR scanner. One quarter of the way through the study the system was upgraded to a GE Signa Twinspeed EXCITE, which included a change in a software platform from v9.0 to v10.0 and a change in a RF array head coil from a 4-channel- to an 8-channel cap coil. To establish data uniformity across this upgrade, a cohort of 32 AGES-RS subjects was scanned twice before the upgrade, and twice after (n⫽128 scans). Objective(s): To demonstrate that CT is reproducible across the scanner upgrade performed in AGES-RS. Methods: CT was measured independently in the 128 scans using the methods described in [1-3]. Statistical analyses were performed both for average cortical thickness as well as for each vertex of the surface models. Statistical significance was assessed using a 5% false discovery rate. Results: There was no change in mean cortical thickness (p⫽0.84) or the variance between scans (p⫽0.81) as a result of the scanner upgrade. When tested at every vertex of the surface models, the only difference in CT was found in the occipital lobe. Conclusions: Reproducibility of computer algorithms is essential to establish in studies of dementia. The measurement of cortical thickness (CT) as described in [2,3] is robust against variations in scanner configurations. The small regional effect found in the occipital lobe in this study is consistent with the location of MRI signal changes that result from differences in receiver coil geometry. Preliminary experiments, results of which will be presented, indicate that this occipital effect can be eliminated by adapting the parameters of the intensity nonuniformity correction algorithm [4] to account for multi-channel coil designs. This study further establishes MR based cortical thickness algorithms to be reproducible and robust to hardware changes [5,6,7].
[1] Zijdenbos et al., IEEE TMI, 21(2), 2002 [2] Lerch et al., NeuroImage 24, 2005 [3] Ad-Dab’bagh et al., Proc. 12th annual meeting OHBM, 2006 [4] Sled et al., IEEE TMI, 17(1), 1998 [5] Lerch et al., Cerebral Cortex 15(7), 2005 [6] Lee et al., NeuroImage 31, 2006 [7] Han et al., NeuroImage 32, 2006
P-105
THREE-DIMENSIONAL RECONSTRUCTION OF HISTOLOGICAL SECTIONS AND CORRELATIONS WITH MRI FROM TRANSGENIC MODELS OF ALZHEIMER’S DISEASE
M. Mallar Chakravarty1, Barry J. Bedell1,2, Simone P. Zehntner1, Kurt Hemmings1, Edith Hamel1, Alan C. Evans1,2, D. Louis Collins1, 1 Montreal Neurological Institute, Montreal, PQ, Canada; 2Neuralyse Inc., Montreal, PQ, Canada. Contact e-mail: [email protected] Background: Non-invasive imaging modalities, such as MRI and PET, show great promise in longitudinal studies of transgenic mouse models of Alzheimer’s disease (AD). The routing use of imaging markers, however, requires extensive validation against gold-standard, ex vivo tissue studies. To this end, we have developed sophisticated methods to allow for fully-automated quantification of histological measures. Objective: In this abstract, we outline one of the building blocks in our fully-automated quantification process. This includes careful acquisition of serial two-dimensional histological slices (2D), the reconstruction of these serial histological slices into a three-dimensional (3D) histological volume, and the mapping of this volume onto an MRI atlas. Methods: Histological Data Acquisition: All APP mice brains were formalin-fixed and embedded in a specially dyed paraffin block to enhance tissue-to-paraffin contrast. Whole-brain, serial coronal histological sections were prepared and stained on an automated tissue stainer with Luxol fast blue/Hematoxylin & Eosin (H&E), for highcontrast of anatomical structures. The stained sections were digitized using a Zeiss MIRAX Scan ultra-high resolution, high-throughput, whole-slide scanner. Additionally a block-face photograph was acquired for each section and acts as a local reference for the histological data. Reconstruction of Histological Data: The reconstruction of the histological data requires a three-step process. In the first step, the digital blockface images were reconstructed as a contiguous 3D volume. The analogous slice of histological data was then transformed to fit the blockface using a three-parameter linear transformation. In the final step, the slice-to-slice morphological inconsistencies were corrected using 2D nonlinear warping techniques. The histological volume was warped to fit onto an MRI atlas created from the unbiased nonlinear average of ten T2-weighted mouse volumes. Results: Figure 1 depicts the reconstructed histological volume and mapping onto the MRI atlas. Note that the crosshairs identify the same anatomical structures on the in vivo and ex vivo volumes. Conclusions: These novel methods which we have developed will allow for automated, robust correlations of in vivo and ex vivo studies, and provide for rigorous validation of new imaging techniques, as well as the seamless integration of molecular, histopathological and structural/functional imaging studies.