Working Memory In Normal Subjects And Schizophrenia Patients: A Functional Magnetic Resonance Imaging Study

Part IDynamic cortical network of parametric digit n-back working memory task in normal subjects: A study about effects of memory load with fMRIObjective: With fMRI and parametric digit n-back working memory task, to explore the neural basis of working memory firstly. Second, within the scope of working memory capacity of high-performancing normal subjects, to explore the neural substrate of unconstrained memory capacity via analyzing load dependent effects of brain activity in working memory related areas. And in relative low-performancing normal subjects, while at working memory load as capacity is exceeded, to analysis what happens to brain activity in working memory related areas, consequently to discuss the physiological basis of constrained memory capacity. Finally, to explore TID brain regions in WM task and its mechanism.Materials and Methods: 35 normal subjects received fMRI with a parametric digit n-back task, n=l, n=2, n=3 task are employed, with a 0-back control condition. Low-load working memory task (i.e. lback task), moderate-load working memory task (i.e. 2back task), and high-load working memory task (i.e. 3back task) wereperformance independently. Reaction time and accuracy were recorded across three different load task respectively, and SPSS 10.0 version was used to perform statistical calculations. A p value of less than 0.05 was viewed as significantly difference. On the basis of behavioral performance of 3back task in experiment, all normal subjects was divided into two groups: HPN0(accuracy exceeding or equal to 85%) and RLPNO(accuracy less than 85%). Preprocessing, statistical analysis and result display of functional data were performed using SPM2. One sample /-tests of basic model were used to perform group analysis. Two-sample /-tests were used to perform inter-group contrast. Statistical analyses included: group effects analysis of n-back task in normal subjects;Group analysis across three different load task in HPNO and RLPNO;Group analysis of Oback-lback contrast to explore TID brain regions. To explore load dependent effects of WM related brain regions and TID regions, lback vs 2back> 2back vs 3back> 3back vs lback inter-group contrast were performed in HPNO and/or RLPNO respectively.Results: Six cortical regions in the parametric digit n-back task were defined: bilateral DLPFC (BA9/46), bilateral VLPFC (BA44/45/47), bilateral LPMA (BA6/8), dorsal cingulate /SMA/MPMA (BA6/8/32), bilateral rostral prefrontal cortex and frontal pole (BA10), bilateral PPC (BA7/40). In addition, bilateral cerebellum, left basal ganglia region and thalamus region also demonstrated activity. In HPNO, across three different load task, left DLPFC, left VLPFC, left LPMA, left MPMA, bilateral PPC showed positive load-dependent activity. In RLPNO, across three different load task, brain areas showing load-dependent activity included: right DLPFC(BA46), left VLPFC(BA45/47), left LPMA (BA6), right superior parietal lobule(BA7) and left precuneus(BA7), left basal ganglia region, SMA/dorsal cingulate(BA6/32). Those load-dependent areas displayed three different physiological response patterns: (l)capacity constrained pattern. The areas included: right DLPFC(BA46), left VLPFC(BA45/47), left basal ganglia region. (2)capacity unconstrained pattern, suchas SMA/dorsal cingulate (BA6/32). (3)capacity independent pattern. Such as LPMA (i.e.. left middle frontal gyrus, BA6), right superior parietal lobule (BA7) and left precuneus (BA7). And at the highest load task(i.e. 3back task) left superior parietal lobule were recruited. For HPNO, task induced deactivation brain regions included: MPFC, cingulate, right inferior frontal gyrus(BA47), multiple regions in bilateral temporal lobe;bilateral postcentral gyrus(BA5). Within the scope of 2back load level, activation of TID brain regions increased with increasing load, and most TID regions showed platform manifestation once the 2back level was exceeded.Conclusion: The neural substrate of the parametric digit n-back task is a distributed brain areas network including cortex, subcortical regions and cerebellar areas. High performance in HPNO has corresponding physiological basis. Within the scope of working memory capacity, as memory increasing parametrically, HPNO mainly depends on left laterality function of brain to implement task accurately. Decreased behavior performance in RLPNO also has physiological basis, with using a WM load which exceeded capacity limit, working memory related brain areas shows constrained capacity changes while others regions play a complementary role. Increasing task load to a certain extent (i.e. 2back load level), TID brain regions also showed increased deactivation while activity of WM related brain areas increasing. So within the scope of 2back load level, ongoing processes of resting state and/or control baseline state and cognition processing state are two extremeties of working memory continuum. Brain resource shifting between the two extremetries is the basis of TID.Part IIDynamic cortical network of parametric digit n-back working memory task in schizophrenia subjects: A study about effects of memory load with fMRIObjective: This article examined neural substrates of working memoryimpairment in schizophrenia with fMRI and parametric digit n-back working memory task through the flowing aspects. Firstly, to analyse WM related brain regions for digit n-back task in schizophrenia subjects. Second, to analyse load dependent effects of brain activity in WM related areas within the scope of WM capacity of high-performancing schizophrenia subjects (HPSCH), and beyond the scope of WM capacity of low-performancing schizophrenia subjects (LPSCH). Third, to compare activity difference of WM network between HPSCH and HPNO, LPSCH and RLPNO at different load levels. Fourth, to identify TID network in HPSCH, its load dependent effects, and to explore difference of load dependent effect between HPSCH and HPNO. Then, we characterize activation difference between HPNO and HPSCH for TID network across variable load levels. Lastly, to analyse the value of WM network and TID network, to verify the feasibility of combing the two networks to differentiate two separate data, one from LPSCH group and the other from LPNO. Materials and Methods: 30 paranoid, undifferentiated schizophrenia subjects and 35 normal subjects received fMRI with a parametric digit n-back task, n=l, n=2, n=3 task are employed, with a 0-back control condition. On the basis of behavioral performance of 2back task in formal experiment, all schizophrenia subjects were divided into two groups: HPSCH(accuracy exceeding or equal to 75%) and LPSCH(accuracy less than 75%). Reaction time and accuracy were recorded across three different load tasks respectively. SPSS 10.0 version was used to perform statistical calculations. Separate Intra-group analysis for HPSCH and LPSCH group was performed, also inter-group analysis for the HPSCH and HPNO, LPSCH and RLPNO. A P value of less than 0.05 was viewed as significantly difference. Preprocessing, statistical analysis and result display of functional data were performed using SPM2. One sample f-tests were used to perform group analysis. Two-sample /-tests were used to perform intergroup contrast. Statistical analyses included: group effects analysis of n-back task in schizophrenia;Group analysisacross three different load task in HPSCH and LPSCH;Group analysis of Oback-lback contrast;To explore load dependent effects of WM related brain regions and TID regions, lback vs 2back, 2back vs 3back, 3back vs lback intergroup contrast were performaned in HPSCH and/or LPSCH respectively. To explore brain functional difference, HPSCH vs HPNO and LPSCH vs RLPNO contrast at three variable load level were also established.Results:(T)Digit n-back WM related regions of schizophrenia is identical to that of NO. ?The load-response function is also consistent to an inverted-U shape. Within the scope of 1,2,3 back load level in HPSCH (i.e. ascending limb of inverted-U shape function), bilateral VLPFC, right DLPFC activation increased parametrically with WM load, but overall WM load sensitivity decreased while compared to HPNO. Moreover, while compared to RLPNO, LPSCH subjects showed a more speeded reach in capacity limit, at the point brain activation demonstrated most prominent, and declined at the higher load as capacity was exceeded(i.e. descending limb of inverted-U shape function). Given the capacity of WM system didn't be exceeded in HPSCH even at the maximum load, activation of WM network is equal to(i.e. at lback level in HPSCH) or beyond the normal level(i.e. cortical function inefficiency at 2back load level in HPSCH and lback load level in LPSCH) when subjects perform above the chance and at matched consistent accuracy. Regardless WM capacity is exceeded or not, decreased performance lead to decreased activation of WM network, i.e. at 3back load level in HPSCH and 2> 3back load level in LPSCH. ?HPSCH consistently engaged TID brain regions similar to those observed in HPNO when performing the n-back task. Contrast to HPNO, the parametric increase of WM load didn't further decrease the deactivation degree, on the contrary lead to invariable deactivation signals and even enhanced signals. At the low, intermediate load level (i.e. l,2back level in HPSCH), while activation of WM network is equal to or beyond the normal level, few TID brain areas show decreaseddeactivation. At the high load level (i.e. 3back level in HPSCH), while activation of WM network in HPSCH decreased obviously, widespread TID brain areas showed decreased deactivation. ?Combing WM network and TID network can differentiate two separate data, one from LPSCH group and the other from LPNO.Conclusion: (D WM impairment in schizophrenia patients has a pathophysiological basis. WM network but not simple brain region was compromised, resulting in decreased WM capacity or increased effective burden imposed by WM task and leading to pathological shift of load-response function leftward. But the left shift of identical magnitude curve in normal subjects is too simple to explain all activation changes in schizophrenia. As a consequence, WM network could show normal or ineffective and increased activation within the scope of capacity limit. Once the ability of ongoing recruiting brain resource decreased, even with the capacity scope, activation of WM network decreased. And while WM capacity was exceeded by the imposed load, WM network shows decreased activation and behavioral impairment. (2) Ineffective resource reallocation is the basis of WM deficits of schizophrenia patients. Disruption or reversion of functional connectivity between prefrontal region and limbic system, perilimbic cortex may be an aspect. Abnormal resource reallocation result in disappearance of the coupling relationship between "resting state and/or control state" and "external task state", the parametric increase of WM load doesn't further decrease the deactivation degree, but leads to invariable deactivation signals and even enhanced signals. (3)WM impairment is the result the concerted interplay of WM network and TID network, and the result of overall functional impairement. ?Schizophrenia subjects and normal subjects have different activation pattern. TID network is specific for diagnosis of schizophrenia subjects group. At different load levels, TID network shows decreased deactivation to a different extent, with most salience while WM network activation decreases obviously. This pattern can be regarded as a trait maker.Part HIPhase specific patterns of cortical activity for working memory in normal subjects: An fMRI studyObjective: This study investigated patterns of cortical activity, and brain regions of load-related activity associated with the temporally separated encoding, maintenance and retrieval phases of working memory (WM) using event-related fMRI (ER-fMRI).Materials and Methods: Twenty healthy subjects completed a relative lower, high load version of the Sternberg Item Recognition Task during ER-fMRI. The brain areas showing activity on average across the two loads for each phase were obtained with group analysis. Load dependent activity for each phase was identified with inter-group analysis.Result: During encoding the contrast for mean activity across the two memory loads revealed significant activation in a distributed network of regions including visual association cortices in the occipital and temporal lobes, right SMA and adjacent regions, bilateral LPMA, left parietal lobe. For high load level, activation of bilateral occipital and temporal area, left LPMA, left SMA increased. In addition, left DLPFC, left primary motor cortex also showed activation. Our findings of significant activation during maintenance phase were in DLPFC, VLPFC, SMA and PMA, visual association cortices. Brain regions including VLPFC, SMA and premotor cortex showed increased activation for high load level. During retrieval, the contrast of mean activity across the two loads levels revealed significant activation in SMA/CMA, MPMA, LPMA, primary motor cortex, DLPFC. Only retrieval related activity in SMA/CMA is load related.Conclusion: ?Distinct but overlapping neural substrates are identified in different cognitive process. ?Encoding is related with visual pattern processing, the storage of visual and verbal information, possibly accompanying with strategic encoding and preparation of a specified motor response. Among the regions, visual association cortices in the occipital and temporal lobes are most specific, which demonstrates increased activation for high load level. ?Maintenance phase is associated with information maintenance, subverbal rehearsal, storage. VLPFC, SMA and PMA responsible for information maintenance, subverbal rehearsal are relatively specific, showing load related increased activation for high load level. Parietal cortex is related to maintenance of verbal information. In addition, DLPFC also displays activity which maybe related to maintenance, manipulation(etc. strategic maintenance), and response preparation. ?Retrieval phase is involved in motor execution and probe-related cognitive processes, resulting in increased SMA/CMA,MPMA, LPMA, primary motor cortex, and DLPFC. ?Encoding and maintenance are associated with extensive load dependent activity, while retrieval is associated with more selective load dependent activity, only in the SMA/CMA, which is responsible for the more abstract, higher order and load dependent aspects of response preparation, such as assembly and selection of movement programs.Part IVPhase specific patterns of cortical activity for working memory in schizophrenia patients: An fMRI studyObjective: To investigate patterns of cortical activity, and brain regions of load-related associated activity associated with the temporally separated encoding, maintenance and retrieval phases of working memory (WM) in schizophreniapatients using event-related fMRI (ER-fMRI), and to explore the relationship between WM deficit of schizophrenia patient and the three cognitive processes.Materials and Methods: Twenty schizophrenia subjects completed a relative lower, high load version of SIRT using fMRI. The brain areas showing activity on average across the two loads for each phase were obtained with group analysis. Load dependent activity for each phase was identified with inter-group analysis. Activity difference between schizophrenia subjects and normal subjects for each phase was also identified with inter-group analysis.Result: During encoding the contrast for mean activity across the two memory loads revealed significant activation in multiple regions, including visual association cortices in the occipital and temporal lobes(BA19/37), bilateral PPC(BA7/40), bilateral LPMA(BA6/8), right cerebellum, and left precentral gyrus(BA4). The majority of above mentioned areas showed enhanced activity in high load level. Compared to normal subjects, aforementioned areas also demonstrated increased activation. During maintenance phase bilateral VLPFC(BA44/45/47), right precuneus lobe(BA7) were activated. For high load level, activation of bilateral PPC(BA7/40), right DLPFC, bilateral occipital lobe(BA17/18/19) increased. And compared to normal subjects, left VLPFC(BA47), bilateral PPC(BA BA7/19/40), right LPMA(BA8), left DLPFC(BA9), bilateral occipital lobe(BA18/19/37) demonstrate increased activation. During retrieval, the contrast of mean activity across the two loads levels revealed significant activation in left precentral gyrus(BA4), left postcentral gyrus(BA3), left SMA(BA6), bilateral LPMA(BA6/8), left DLPFC(BA9). For high load level, activation of right LPMA(BA8), left precentral gyrus(BA4/44), right cerebellum, right DLPFC(BA46), bilateral occipital lobe(BA18/19/37) increased. Compared to normal subjects, left LPMA(BA6), bilateral cerebellum, bilateral precentral gyrus(BA4), left precentral gyrus(BA44), left postcentral gyrus(BAl/5^ left DLPFC (BA9) displayed increased activation. Inencoding, maintenance and retrieval phases, emotion processing related brain regions and resting state related areas showed increased activation in schizophrenia subjects, especially in the encoding phase.Conclusion: ?Averaged activities across encoding, maintenance and retrieval phases of working memory in schizophrenia group are generally similar to those of NO group, suggesting schizophrenia patients recruit the same brain regions to subserve WM performance. There isn't significant difference of load dependent activity between groups during encoding phase. But schizophrenia subjects show a lesser extent of load dependence than normal subjects for maintenance phase. And for retrieval phase the schizophrenia group shows a significant load dependence of brain regions that involved in lower order motor function preparation and execution, but not SMA/CMA. (2)WM deficits isn't the result of specific cognition process impairment, but the coordinated result of the abnormity of encoding, maintenance and retrieval phase. Within the scope of capacity, brain regions show ineffective, increased activity across the three phases given matched accuracy of performance and populations characteristics between the two groups. ?Schizophrenia patients couldn't optimize resources utilities, but recruit task-irrelative brain regions, suggesting coupling and competition relationship between "resting state" and "cognition state" disappear in schizophrenia patients.