| Abstract This research demonstrates the feasibility of using functional MRI (fMRI) to investigate cognitive deficits in childhood cancer survivors. Because of possible cerebral vasculature damage due to tumor or therapy, it was unknown whether fMRI techniques were applicable to this population. This dissertation hypothesized that (1) childhood cancer survivors have normal hemodynamic responses to visual stimulation, and (2) blood oxygenation level dependent (BOLD) fMRI is a useful tool to investigate cognitive deficits in this population. A brief visual stimulation fMRI paradigm was implemented to measure the hemodynamic response function (HRF), and an attention task fMRI paradigm based on the Continuous Performance Task (CPT) was designed to map the brain activation during cognitive processes. The study population comprised 31 healthy volunteers and 16 survivors. For the visual stimulation, survivors showed normal primary visual cortex activation, albeit a smaller activation size than did volunteers. The measured visual HRF for survivors was qualitatively normal in timing and shape; but their BOLD signal showed a lower peak of the positive response and a larger post-stimulus undershoot than volunteers'. Brain areas activated during the CPT for volunteers included major components of attention networks; whereas activated areas for survivors had both common and different areas. These differences may relate to cognitive deficits, different cognitive strategies, or alterations in HRFs of survivors. To investigate the last possibility, measured HRFs of seven survivors were used in CPT fMRI data analyses. The returned results were comparable to those of the method using the canonical HRF; and this similarity suggests that BOLD responses across brain regions are qualitatively normal in survivors, and fMRI is a useful tool to investigate their cognitive deficits. However, in two cases, brain areas associated with attention functions were identified as activated by using subject-specified HRF but as not activated by using the canonical HRF, and vice versa. These discrepancies indicate that activation may not be detected simply because of alterations in survivors' HRFs. Therefore, inclusion of subject-specified HRFs in data analysis may facilitate the interpretation of fMRI results. In light of the finding that their HRFs are basically normal, a wide range of research is now possible with the use of fMRI in childhood cancer survivors. |
