{"id":{"repo_id":"njit","oai_identifier":"oai:digitalcommons.njit.edu:theses-1327"},"canonical_url":"https://search.dev.ndltd.org/etd/njit/oai:digitalcommons.njit.edu:theses-1327","repository":{"repo_id":"njit","name":"NJIT","base_url":"https://digitalcommons.njit.edu/do/oai/"},"display":{"title":"Comparison of linear parametric models for predicting fMRI response","abstract":"In this study, five different linear parametric models including Autoregressive model (ARX), Autoregressive Moving Average Model (ARMAX), Box-Jenkins Model (BJ), Instrument Variable Model (IV) and Prediction Error Model (PEM) were used to predict the fMRI response and their performances compared. Transfer functions were computed for every voxel time series for every subject using all the parametric models. Cross-correlation was subsequently performed between the predicted response and the actual fMRI data to compare the performance of the five models. The consistency of the models and the transfer function was checked by doing a statistical analysis. Among the five models tested, PEM resulted in the highest correlation coefficient of 0.76 with the measured response, while ARX, which was the simplest of all, gave the least correlation coefficient of 0.23 with the measured response. The PEM model was consistent in predicting the response between the subjects compared to all other models. A significant difference between the PEM model versus the other models was observed for all the subjects.","abstract_html":"In this study, five different linear parametric models including Autoregressive model (ARX), Autoregressive Moving Average Model (ARMAX), Box-Jenkins Model (BJ), Instrument Variable Model (IV) and Prediction Error Model (PEM) were used to predict the fMRI response and their performances compared. Transfer functions were computed for every voxel time series for every subject using all the parametric models. Cross-correlation was subsequently performed between the predicted response and the actual fMRI data to compare the performance of the five models. The consistency of the models and the transfer function was checked by doing a statistical analysis. Among the five models tested, PEM resulted in the highest correlation coefficient of 0.76 with the measured response, while ARX, which was the simplest of all, gave the least correlation coefficient of 0.23 with the measured response. The PEM model was consistent in predicting the response between the subjects compared to all other models. A significant difference between the PEM model versus the other models was observed for all the subjects.","abstract_has_math":false,"creators":["Gandhi, Parina"],"institution":null,"degree_name":"Master of Science in Biomedical Engineering - (M.S.)","degree_level":null,"degree_discipline":"Biomedical Engineering","degree_department":null,"school":null,"contributors":["Tara L. Alvarez","Bharat Biswal","Max Roman"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2008,"date_issued":"2008-01-27T08:00:00Z","date_published":"2008-01-27T08:00:00Z","updated_at":"2026-07-24T03:23:16Z","subjects":["Parametric model","Hemodynamic response","Autoregressive","Transfer function cross-correlation","Biomedical Engineering and Bioengineering"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://digitalcommons.njit.edu/theses/328","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Tara L. Alvarez","Bharat Biswal","Max Roman"]},{"key":"dc:creator","label":"Author","values":["Gandhi, Parina"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Biomedical Engineering"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science in Biomedical Engineering - (M.S.)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Parametric model","Hemodynamic response","Autoregressive","Transfer function cross-correlation","Biomedical Engineering and Bioengineering"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.njit.edu/theses/328"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["In this study, five different linear parametric models including Autoregressive model (ARX), Autoregressive Moving Average Model (ARMAX), Box-Jenkins Model (BJ), Instrument Variable Model (IV) and Prediction Error Model (PEM) were used to predict the fMRI response and their performances compared. Transfer functions were computed for every voxel time series for every subject using all the parametric models. Cross-correlation was subsequently performed between the predicted response and the actual fMRI data to compare the performance of the five models. The consistency of the models and the transfer function was checked by doing a statistical analysis. Among the five models tested, PEM resulted in the highest correlation coefficient of 0.76 with the measured response, while ARX, which was the simplest of all, gave the least correlation coefficient of 0.23 with the measured response. The PEM model was consistent in predicting the response between the subjects compared to all other models. A significant difference between the PEM model versus the other models was observed for all the subjects."]},{"key":"dc:title","label":"Title","values":["Comparison of linear parametric models for predicting fMRI response"]}]}],"canonical_facts":{"dc:contributor":["Tara L. Alvarez","Bharat Biswal","Max Roman"],"dc:creator":["Gandhi, Parina"],"dc:description.abstract":["In this study, five different linear parametric models including Autoregressive model (ARX), Autoregressive Moving Average Model (ARMAX), Box-Jenkins Model (BJ), Instrument Variable Model (IV) and Prediction Error Model (PEM) were used to predict the fMRI response and their performances compared. Transfer functions were computed for every voxel time series for every subject using all the parametric models. Cross-correlation was subsequently performed between the predicted response and the actual fMRI data to compare the performance of the five models. The consistency of the models and the transfer function was checked by doing a statistical analysis. Among the five models tested, PEM resulted in the highest correlation coefficient of 0.76 with the measured response, while ARX, which was the simplest of all, gave the least correlation coefficient of 0.23 with the measured response. The PEM model was consistent in predicting the response between the subjects compared to all other models. A significant difference between the PEM model versus the other models was observed for all the subjects."],"dc:identifier":["https://digitalcommons.njit.edu/theses/328"],"dc:subject":["Parametric model","Hemodynamic response","Autoregressive","Transfer function cross-correlation","Biomedical Engineering and Bioengineering"],"dc:title":["Comparison of linear parametric models for predicting fMRI response"],"dc:type":["Thesis"],"thesis:degree_discipline":["Biomedical Engineering"],"thesis:degree_name":["Master of Science in Biomedical Engineering - (M.S.)"]},"updated_at":"2026-07-24T03:23:16Z"}