LONI: Laboratory of Neuro Imaging

Diffusion Tensor Imaging

Following an extensive period of testing protocols that focused on the number of directions of data acquisition required, number of repetitions in order to maximize signal to noise properties in the images, and availability of pulse sequence algorithms, we finalized a protocol for the study of probabilisitic white matter tracts and have begun research using this protocol at all participating sites. In order to begin this research we had to obtain pulse sequence code from a variety of different sources, and install and implement it on different platforms and different scanners across the consortium. We are anticipating exciting results from the analysis of white matter tracts in the human brain in a large population and the comparison of these results with post mortem microscopic white matter analyses using differential myelin stains.

Diffusion Tensor Imaging

The mission of this project is to establish a common DTI protocol and to test the quality of DTI data from participating sites. This is an important first step toward generation of a large normal DTI database and probabilistic white matter atlas. After drafting a first version of the protocol we found that:

  1. Gradient orientation tables are not the same.
  2. Imaging matrix of 96x96 can't be performed.
  3. The number of non-diffusion weighted images can't be controlled.
  4. Some scanners have EPI-related Nyqest ghosting.

These issues were due to scanner differences. Based on this early research we developed a second version of the protocol in which the same gradient table, imaging matrix, and the number of non-diffusion weighted images were used for all scanners. We implemented the necessary modifications at all five participating sites and completed the first round of data acquisition (insert link to description and images here).

We are currently in the process of measuring the signal-to-noise ratio of DTI

datasets from the five participating sites to ensure that all sites can produce comparable images. Those institutes that pass this test will start generating normal data sets. This project will then move to the next phase, in which we will create the probabilistic map of white matter anatomy.

Establishment of Diffusion Tensor Imaging Database

ICBM is collaborating with MITRE on developing database software that will be needed for the creation of the DTI human white matter database. This software will have the following role:

  1. This software will allow participating sites to send their DTI data to Johns Hopkins University via web-browser.
  2. This software will automatically perform co-registration of all diffusion-weighted images prior to tensor calculation.
  3. This software will automatically perform the tensor calculation.

We have implemented the first version of the software that allows sites to send their DTI data to Johns Hopkins, and performs co-registration of the diffusion-weighted images.

Comparison Between DTI- and Histology-Based White Matter Delineation

In collaborative work between Johns Hopkins University (JHU) and Institute of Medicince, Jülich (IMJ), Johns Hopkins University is creating the white matter tract atlas and determining the coordinates in a standardized reference space. IMJ is reconstructing the same tracts using histology-based techniques. This enables us to perform a validation study of DTI-based tract reconstruction by comparing these two types of data.

Creating the Probabilistic White Matter Atlas

Johns Hopkins University (JHU) has a DTI database of normal subjects acquired under the Human Brain Project (RO1 AG10012). JHU is collaborating with the Montreal Neurological Institute group to generate a probabilistic atlas in MNI-ICBM and JHU DTI coordinates. We have created our first atlas using a simple linear transformation using the AIR algorithm. An example of a 15-subject average map in the JHU DTI coordinate is shown in Figure 3 (link here).