Characteristic Analysis Of Cognitive Impairment Based On Multimodal Data

Cognitive function is a high-level psychological function,including attention,memory,language,perception and so on.Cognitive impairment refers to the impairment of one or more of the above cognitive functions.Patients with cognitive impairment may have trouble remembering,learning new things,concentrating,or making decisions that affect everyday life.Cognitive impairment will cause patients to lose certain life and social skills,greatly reduce their life quality,and increase the burden on the families and society.Therefore,early screening and intervention for cognitive impairment is crucial.However,the current clinical screening of cognitive impairment is still based on subjective assessment tools and neuropsychological cognitive assessment tools,such as clinical interviews,questionnaires and scales,lacking precise and specific assessment indicators and techniques.With the development of cognitive science study,researchers have begun to use neuroimaging technology to assist the screening of cognitive function,improving the accuracy and objectivity of the assessment.Due to the complex mechanisms of cognitive impairment,it is generally difficult to characterize the cognitive impairment using unimodal data.This dissertation focuses on schizophrenia and temporal lobe epilepsy.A multimodal data acquisition system is used to monitor synchronously the participants’ electrophysiological data and eye movement information during cognitive tasks.The patient’s specific electrophysiological and visual behavioral response patterns serve as an objective and quantitative characteristic index of cognitive impairment.Further,through the study of the characteristics of cognitive impairment,the mechanism of cognitive impairment caused by the disease will be revealed.The main research work of this dissertation is summarized as follows:(1)This is the first study in schizophrenia to monitor simultaneously the hemodynamic response of the prefrontal cortex using functional near-infrared spectroscopy technology during the MATRICS Consensus Cognitive Battery.The relative changes of oxyhemoglobin concentration in the prefrontal cortex during the test of 95 participants are used as monitoring indicators.Our results find that schizophrenia patients score lower than healthy controls on all three tests(p values < 0.001),but show significantly higher activation of prefrontal cortical during the test(p < 0.03).Abnormal activation of the prefrontal cortex reflects abnormal information processing in schizophrenia and can be used to differentiate schizophrenia from healthy controls.Using the prefrontal cortical oxygenation during the three tests to classify schizophrenia and healthy controls,the highest classification accuracy is 94.3%,which further confirms that the prefrontal cortical oxygenation during the task state has the possibility the executive impairment characteristics of schizophrenia.(2)For a more objective and sensitive memory assessment,this dissertation creates a memory assessment platform based on eye tracking technology,which solves the problem of confusion between attention and memory processes in visual memory tasks by synchronizing eye tracking in the short-term memory task.Scale scores,EEG and eye tracking data of 34 temporal lobe epilepsy(TLE)patients and 28 healthy controls are collected to construct a dataset.We define eye tracking statistical indicators,including total visit time,total visit counts,average visit time,and first fixation time to quantify visual behavior.TLE patients take longer to complete the task(p < 0.001)and longer to first fixate on the target(p = 0.049)during the memory decoding phase,and have significantly more visit counts(p = 0.008).The dissertation reveals abnormal memory and attention patterns in temporal lobe epilepsy,that is,patients with temporal lobe epilepsy show impaired memory retrieval,but visual attention is preserved in the short-term memory task.It is proposed that the number of epileptiform discharges in the temporal lobe during sleep is significantly negatively correlated with the performance of the short-term memory(correlation coefficient r =-0.463,p = 0.013),providing more electrophysiological evidence for memory impairment in temporal lobe epilepsy.Compared with the traditional memory scale,the results of the automatic memory assessment platform can reflect more information about the memory impairment of temporal lobe epilepsy and help to reveal the memory impairment mechanism of temporal lobe epilepsy.(3)Extending the previous work,we further explore the differences in memory impairment patterns between the subtype of mesial temporal lobe epilepsy(MTLE)by the memory assessment platform.A dataset is constructed by collecting eye tracking data and Wechsler memory scale scores from 16 MTLE patients with hippocampal sclerosis(HS-MTLE),25 MTLE patients with no lesions on MRI(MRI-neg MTLE),and 35 healthy controls.Both MRI-neg MTLE and HS-MTLE show memory deficits in the completion time of the eye tracking task(p < 0.05),which is consistent with traditional memory assessments.Among the four difficulty levels of the short-term memory task,the MRI-neg MTLE patients have significantly less visit time on the target during the memory encoding phase than healthy controls(p < 0.02),and significantly lower percentage of total visit time on the visual stimuli on the screen during the memory decoding phase than healthy controls(p < 0.02).However,there is no difference between HS-MTLE and healthy controls in the above eye tracking indicators.This dissertation proposes different patterns of memory impairment in two subtypes of MTLE,namely,MRI-neg MTLE show impaired memory with marked attention deficit,whereas HS-MTLE patients show only memory impairment and no attention deficit.The eye-tracking-based memory assessment platform shows a more sensitive assessment ability than traditional memory scales,and is expected to play an important role in the early diagnosis and intervention of neurological disease-related cognitive impairments.