Toolbox for automatic localization of volume of interest in MRS (ALLVOI)
Po-Yu Peng1, Shang-Yueh Tsai2,3, and Yi-Ru Lin1

1Electronic and Computer Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan, 2Graduate Institute of Applied Physics, National Chengchi University, Taipei, Taiwan, 3Reasearch Center for Mind,Brain and Learning, National Chengchi University, Taipei, Taiwan

Synopsis

Magnetic Resonance Spectroscopy (MRS) data were usually acquired with single voxel spectroscopy techniques (SVS), which is used to access metabolites concentrations from a predefined volume of interest (VOI). Currently, MRS has been linked to fMRI studies to access metabolic information. Therefore, it’s important to define the VOI on standard space (template) and transform the predefined VOI from standard space to subject space for each MRS scan. In this study, we developed a tool to guide the definition of VOI on template before MRS scan then automatically calculate VOI in subject space.

Purpose

Magnetic resonance spectroscopy (MRS) is a non-invasive technique to access metabolic information in the brain. Single voxel spectroscopy (SVS) is most common MRS method used to access metabolites concentrations from a predefined volume of interest (VOI). Currently, SVS method has been linked to fMRI studies to access metabolic information in activated brain regions or in resting brain areas.1-3 Therefore, it is important to define the VOI on standard space (template) and transform the predefined VOI from standard space to the subject space for each MRS scan. In this way, VOI can be determined from the template based on a cluster in fMRI results or a known brain structure instead of determining VOI subject-by-subject by operator’s experience. Metabolic information can be directly linked to fMRI results and operator bias on the localization of VOI can be minimized. In this study, an automatic tool is developed for this purpose. A graphical user interface (GUI) is used to guide the definition of VOI on the template before MRS scan. During MRS scan, the coordinate of predefined VOI on standard space can be automatic transformed to subject coordinate in the system/subject using an algorithm developed previously 4.

Methods

The toolbox can be separate into two parts. The flow chart of toolbox is shown in figure 1.

1). User-defined VOI: In this part, a GUI interface is used to display the template in 3 orientation view. Users can define the VOI by entering image coordinate and voxel size. VOI can be tilted and rotated to fit in the brain structure. Users can also load the fMRI results as guidance to locate VOI. Finally, the coordinate of user-defined VOI is saved as a text file in the database.

2). Automatic localization of user-defined VOI: Before MRS scans, a high resolution T1 images, such as MPRAGE, must be acquired and imaging transformation between standard space and system space of each subject can be done by SPM8 and an in-house developed MATLAB script.4 Then, the user-defined VOI (in user-defined text file) can be transformed into system space in console format. Finally, the transformed user-defined VOI is displayed on the GUI interface on the T1 images.

Results

An example of GUI of user-defined VOI is shown in figure 2(a). The MNI template is load as background image and an fMRI result is overlaid on the template. VOI can be defined by adjusting parameters including voxel center, voxel size and orientation. The user-defined VOI is shown in figure 2(b). An example of transformation of this user-defined VOI into subject space is shown in figure 3. The user-defined VOI is transformed to subject space and overlaid on the subject’s T1 image (MPRAGE). We can clearly see that the automatic transformation algorithm can successfully transform user-defined VOI in standard space to subject space for MRS scan. The procedures of automatic location of user-defined VOI take around 3 minutes, which include time for segmentation/normalization in SPM8 and transformation calculation in in-house MATLAB scripts.

Discussuion

Here, we have developed a toolbox to automatically locate the VOI of MRS scan based on user-defined VOI on the template. For combination of MRS and fMRI researches, it is convenient to define the MRS VOI in standard space, in which the brain area can match the fMRI finding. The requirement of this method is the high resolution T1 images used to access the transformation matrix between standard space and subject space. T1 images are already a routine protocol in most MRI studies. Further, the duration in calculating user-defined VOI in subject space take around 3 minutes. Other MRI scans can be incorporated during this period to save scan time.

Acknowledgements

This study was supported in part by grants from the Ministry of Science and Technology (MOST 103-2221-E-011-006-MY3). The authors thank Taiwan Brain and Mind Imaging Center (TMBIC) and National Cheng-Chi University for consultation and instrument availability for this work.

References

1. Wiebking C, et al. GABA in the insula – a predictor of the neural response to interoceptive awareness. Neuroimage 2014 Feb; 86: 10-18.

2. Greicius MD, et al. Default-mode network activity distinguishes Alzheimer’s disease from healthy aging: evidence from functional MRI. PNAS U.S.A. 2004 Mar; 101(13): 4637-4642.

3. Gusnard DA, et al. Medial prefrontal cortex and self-referential mental activity: relation to a default mode of brain function. PNAS U.S.A. 2011; 98(7): 4259-4264.

4. Gong YJ, et al. Automatic position of MR spectroscopy voxels. OHBM 2015 June; Poster Number: 1904.

Figures

Figure 1: The flow chart of ALLVOI.

Figure 2: An example of GUI for user-defined VOI. (a) fMRI result was overlaid on the MNI template. (b) VOI was defined based on the fMRI activation map.

Figure 3: User-defined VOI was transformed into subject space automatically and overlaid on T1 images.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4009