In response to the replication crisis, neuroimaging research has increasingly been moving towards a more collaborative model, with groups around the world collaborating and combining datasets to achieve greater statistical power and identify generalizable results. While promising, this has brought to light new issues regarding how groups can meaningfully combine data collected with different parameters. Differences in scanner platforms and scan parameters can introduce variations in image intensity, leading to alterations in the extracted measures. Along with strategies for harmonizing data, researchers across the field are determining how to establish the validity of “harmonized” data when there is no available ground truth.
With this Research Topic, we would like to define “harmonization” in this context and discuss factors that researchers must examine when considering harmonization of neuroimaging data. Before embarking on such an effort, researchers must be aware of why harmonization is important, when it is appropriate and when it is not, potential strategies for harmonization and the limitations of these, and how to assess whether output data are valid. We hope that this Topic will serve as a resource for the community and provide a forum for discussing related issues. This Topic will present the current state of the field and discuss best practices going forward as the field continues to develop rapidly.
Therefore we welcome the following manuscript submissions:
• Empirical studies testing harmonization approaches of imaging data across sites/platforms/vendors and/or time, for any modality commonly used for neuroimaging, including:
o Functional MRI/resting state fMRI
o Diffusion MRI
o PET
o MRS
o Structural MRI/volumetrics
• Commentary on the importance of, recent developments in, or pitfalls with image harmonization
• Theoretical review of how to validate “harmonized” data
• Use of phantom objects or mathematical algorithms to adjust data
• Position papers on “best practices” in image harmonization
• Research integrating harmonization across imaging and non-imaging variables (e.g. clinical or demographic variables)
Topic Editor Maxime Descoteaux is a co-founder, shareholder, and the Chief Science Officer at Imeka. The other Topic Editors declare no competing interests with regard to the Research Topic subject.
In response to the replication crisis, neuroimaging research has increasingly been moving towards a more collaborative model, with groups around the world collaborating and combining datasets to achieve greater statistical power and identify generalizable results. While promising, this has brought to light new issues regarding how groups can meaningfully combine data collected with different parameters. Differences in scanner platforms and scan parameters can introduce variations in image intensity, leading to alterations in the extracted measures. Along with strategies for harmonizing data, researchers across the field are determining how to establish the validity of “harmonized” data when there is no available ground truth.
With this Research Topic, we would like to define “harmonization” in this context and discuss factors that researchers must examine when considering harmonization of neuroimaging data. Before embarking on such an effort, researchers must be aware of why harmonization is important, when it is appropriate and when it is not, potential strategies for harmonization and the limitations of these, and how to assess whether output data are valid. We hope that this Topic will serve as a resource for the community and provide a forum for discussing related issues. This Topic will present the current state of the field and discuss best practices going forward as the field continues to develop rapidly.
Therefore we welcome the following manuscript submissions:
• Empirical studies testing harmonization approaches of imaging data across sites/platforms/vendors and/or time, for any modality commonly used for neuroimaging, including:
o Functional MRI/resting state fMRI
o Diffusion MRI
o PET
o MRS
o Structural MRI/volumetrics
• Commentary on the importance of, recent developments in, or pitfalls with image harmonization
• Theoretical review of how to validate “harmonized” data
• Use of phantom objects or mathematical algorithms to adjust data
• Position papers on “best practices” in image harmonization
• Research integrating harmonization across imaging and non-imaging variables (e.g. clinical or demographic variables)
Topic Editor Maxime Descoteaux is a co-founder, shareholder, and the Chief Science Officer at Imeka. The other Topic Editors declare no competing interests with regard to the Research Topic subject.