The team that created iBEAT recently released iBEAT V2.0 Cloud (https://ibeat.wildapricot.org). This service allows researchers to upload anonymized T1- and T2-weighted images. The iBEAT team will segment the images into gray matter, white matter and CSF, and they can provide additional processing (e.g., atlas mapping, cortical thickness calculations) upon request.

Our neuroimaging team at the Children’s Learning Institute has uploaded 35 toddler MRIs to iBEAT Cloud to be processed. We have been able to receive the segmentation output within about 48 hours after uploading, and we have been thoroughly impressed with the quality of the segmentation. 

In order to use iBEAT Cloud, you need to create a free account on their website. You should also anonymize your MRI data. I changed all subject IDs before uploading to iBEAT Cloud as an added layer of protection (e.g., SubjectXYZ123 became subject-1, etc.). I grouped subjects by age, as the upload sheet asks for the age of your subjects in months. iBEAT Cloud allows each research team to upload 10 subjects per day, so I uploaded our data over several days. For example, on day 1, I uploaded all of our 14-month-old subjects; on day 2 I uploaded all of our 16-month-old subjects; on day 3 I uploaded all of our 21-month-old subjects; etc. In order to upload our data from 35 toddlers, it took me six days of uploading the data like this. 

When I received the segmentation output, I checked each subject the way I would if I were processing the data myself. I edited the segmentation output when necessary. Our subjects are not typically-developing toddlers, so segmentation can be more difficult than with typically-developing samples. We made minor edits to about 20% of the subjects – these were predominantly subjects who were under 18-months-old for whom we only had a T1-weighted image and not a T2-weighted image. The iBEAT team recommends uploading both a T1- and T2-weighted image for subjects under 18-months-old, so we expect that the lack of a T2-weighted image caused these minor segmentation issues. In another 5% of subjects (2 of 35) we made larger edits to the segmentation output. We sent our edits back to the iBEAT team as they requested in order to aid in future segmentation.

Throughout this process, the iBEAT team, especially Dr. Li Wang, has been very helpful and responsive to my questions. They truly understand what the difficulties are with infant and toddler neuroimaging, and they are working hard to produce a product that makes these projects easier for researchers. Ultimately, I expect that iBEAT Cloud will allow us to focus on answering our research questions rather than creating lab-specific methods for processing the toddler MRI data we have collected. I encourage anyone who has struggled with infant, toddler, or preschool MRI data to consider uploading your data to iBEAT Cloud, rather than trying to solve analysis problems on your own.