Validation of spatial normalization of the brain diffusion tensor imaging (DTI) calculated by statistical parametric mapping (SPM)

2002 Fiscal Year Final Research Report Summary

• Project/Area Number: 13670909
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Radiation science
• Research Institution: Tohoku University
• Principal Investigator: HIGANO Shuichi Tohoku University Hospital Research Associate, 医学部附属病院, 助手 (20173148)
• Co-Investigator(Kenkyū-buntansha): MUGIKURA Shunji Tohoku University Hospital Medical Staff, 医学部附属病院, 医員 ; TAMURA Hajime Tohoku University Hospital Research Associate, 医学部附属病院, 助手 (20333817) ; TAKAHASHI Shoki Tohoku University Graduate School of Medicine Professor, 大学院・医学系研究科, 教授 (80148874)
• Project Period (FY): 2001 – 2002
• Keywords: brain / magnetic resonance imaging / diffusion tensor imaging / normalization / atatical parametric mapping

Research Abstract

Purpose : To validate the accuracy and reproducibility of the spacial normalization of DTI. Materials and methods
Subjects include one healthy volunteer, 9 patients without obvious MR abnormality and 6 cases with chronic cereberovascular disease. DTI was obtained using SE type single shot EPI sequence with 6 different direction of motion probing gradients (MPG, b=1000 sec/mm2). In the volunteer, DTI was performed with 6 different imaging parameters. The DTI data were normalized using SPM 99, followed by calculating apparent diffusion coefficient (ADC) images and fractional anisotropy (FA) images. To validate the accuracy of normalization, mismatch of various anatomical landmarks such as cerebral margins, central sulcus, margin of the corpus callosum, Monro foramen, pineal region, center of the middle cerebellar peduncle was measured among cases. To see whether the quantitativeness of ADC and FA was preserved after normalization, correlation of these values between raw images and normali zed images was evaluated by defining regions of interest in various brain structures including middle cerebellar peduncle, cerebral peduncle, basal ganglia, thalamus, internal capsule, corpus callosum and centrum semiovale.
Results Mismatch of the landmarks among the same cases with different studies was 1.2 mm in average (maximum 8mm). The mismatch among different cases was 2.8 mm in average (max 15mm). The difference between the same cases and the different cases was statistically significant (Mann-Whitney: p<.0001). There was good correlation of ADC and FA values in various structures between raw images and normalized images (r=0.992 in ADC, r=0.989 in FA). The correlation is almost identical in ADC but slightly underestimated in FA.
Conclusion Using SPM 99, DTI was well normalized anatomically with preserving quantitativeness, Application of normalization should provide reliable and subjective results in analysis of DTI data obtained from various size/shape of brain with different scan parameters.