October 05, 2015

Xcode builder error when installing SPM on Mac

When the software Statistical Parametric Mapping (SPM) is installed on the Mac platform, the app Xcode is required to compile mex files. A basic guide for installation is introduced in the following Wiki page. You might suffer from the following error
xcodebuild: error: SDK "macosx10.7" cannot be located.
It may be caused by technical incompatibility between Xcode and SPM. To solve this problem, the following customization would be helpful.

Step 1. SPM option

To let SPM use the correct mex functions, put the full path at the term MEXBIN in the file /src/makefile.var as follows as an example
MEXBIN = /Applications/MATLAB_R2012a.app/bin/mex

Step 2. MATLAB option

Go to the SDK directory in Xcode to check the SDK version as introduced in this discussion; for example,
Then you would see the files such as "MacOSX10.xx.sdk". Open MATLAB, and execute the following command
mex -setup
then select "1" so that a user-specific options file for building MEX-files named "mexopts.sh" is copied to a specified folder (shown in the command window) from the template option file. Open the file "mexopts.sh", and replace the original version (probably four texts) to updated version verified from the SDK directory; for example "10.7" to "10.9". After saving the file, follow the SPM installation process again. (For additional information, refer to the discussion).

March 02, 2015

A Talk on Fetal fMRI at SPIE Medical Imaging

I had an oral presentation at the SPIE Medical imaging conference which held in Orlando, Florida at February 25, 2015. At the fMRI session of the conference on Biomedical Applications in Molecular, Structural, and Functional Imaging, the presented topic was about our proposed preprocessing pipeline for handling severe fetal movement effectively in functional MRI of the fetal brain and placenta during maternal hyperoxia (see the figure).

An example of fMRI image including the fetal brain and placenta
One of the key strategies in the proposed method is to optimize each preprocessing step to the experimental design by processing different phases and ROIs of acquisition separately. In spite of these advanced motion correction strategies, fetal motion is not perfectly corrected. To overcome the practical limitation, we also suggested the volume outlier rejection method which automatically excludes high-motion volumes and all the missing data in ROI-averaged signals are imputed by regression. We showed that our method was effective to acquire reliable ROI-averaged time series from BOLD signals disturbed by severe fetal motion without large data loss, comparing to traditional preprocessing methods.

There were some questions regarding volume outlier rejection and ROI definition, which stimulated me to improve them through the future works. The paper was published in the SPIE proceedings on medical imaging, and can be downloaded from the SPIE Digital Library.
Wonsang You ; Ahmed Serag ; Iordanis E. Evangelou ; Nickie Andescavage ; Catherine Limperopoulos; Robust motion correction and outlier rejection of in vivo functional MR images of the fetal brain and placenta during maternal hyperoxia. Proc. SPIE 9417, Medical Imaging 2015: Biomedical Applications in Molecular, Structural, and Functional Imaging, 94170O (March 17, 2015); doi:10.1117/12.2082451.

January 31, 2015

Why do I study the fetal brain?

I am a computational neuroscientist who studies the brain functions of the fetus using functional magnetic resonance imaging (fMRI). Just one year passed since I was first involved in a research project pertinent to pediatric neuroimaging. In spite of short-term experience, I have fallen in fervent love with the fetus. Indeed, it has extraordinary attraction compared to the adult brain.
Its anatomy and functions evolve in every moment as it grows inside the mother’s womb. Imagine how dynamic the change in the fetal brain is. As lots of neurons are newly formed everyday, the fetus extends its interaction with external environment. It has been well known that some classical music can be used as the prenatal care, which means that the fetal brain responds to external sounds. The dynamic evolution of the fetal brain is so mysterious that it sometimes reminds me of the creative providence of the God. All the mysteries lead me to get hooked on the research of the fetal brain. Although my background is originally rooted on engineering and computer science, I have great curiosities on the biological nature of the fetus.
However, the current knowledge on the fetal brain is extremely insufficient. The physical mechanism of fetal brain development is still unclear, and we still have lots of unsolved questions. How does the functional network of the fetal brain evolve over ages? How does the placenta influence the fetal brain functions? What physiological disorders disturb the natural development of the fetal brain functions? One of the main reasons which make it difficult to answer such questions is the technical limitation. The fetal brain should be definitely investigated in a non-invasive manner. In these reasons, the neuroimaging techniques based on magnetic resonance have been preferred by radiologist and neuroscientists working on pediatrics. However, since the fetus moves spontaneously and its motion is out of control, its imaging data get much erroneous.
As the technical limitations are coped with, more scientific facts would be revealed by pediatric radiologists and neuroscientists, and the mysteries of the fetal brain would get unveiled in the near future. I dream forward to meeting such a day.

October 05, 2012

Presentation at Resting State Conference

I presented an abstract titled "Fractal-driven distortion of resting state functional networks in fMRI: a simulation study" at the 3rd Biennial Conference on Resting State Brain Connectivity which took place in Magdeburg, Germany during September 5~7, 2012. The main idea of my presentation is to propose an extended model of hemodynamic response function (HRF) to describe fractal behavior which has been observed in resting state BOLD signals. Refer to this page for detailed information on this abstract.

November 18, 2011

LiveVideo: Real-time Video Analysis Software

The LiveVideo, an open-source real-time object detection and tracking software for H.264/AVC videos, was released. This software was developed by Wonsang You at the MCCB Lab in KAIST at 2007. It enables us to effectively detect and track multiple objects in real time in H.264|AVC bitstreams with static background by applying probabilistic spatiotemporal macroblock filtering (PSMF) and partial decoding technique. This software exhibits reliable performance in a variety of natural scenes with static background as well as fast computation time (30~50 frames/sec) enough to be performed in real-time.

You can freely download its source codes from the Sourceforge project. To quickly study the usage of this software and to understand the structure of source codes, you can refer to the Tutorial document. To figure out theoretical backgrounds, visit the homepage of the corresponding research project and read its relevant paper. Also, the Wiki site of LiveVideo will provide you useful information. If you need technical supports and helps from the software developers, you can either use the help forum or subscribe the mailing list called "livevideo-help". This software is free, but whenever anyone use this software for his/her publications, he/she should cite it as follows.
Wonsang You, M.S. Houari Sabirin, and Munchurl Kim, "Real-time detection and tracking of multiple objects with partial decoding in H.264/AVC bitstream domain," Proceedings of SPIE, San Jose, CA, USA: SPIE, 2009, pp. 72440D-72440D-12.
We seek developers who want to contribute to the LiveVideo open-source project.  Any member may contribute to either improving the performance of object detection and tracking algorithm through theoretical study, or improving the user interface and efficiency of source codes. The collaborative works also include publishing technical articles with co-authorship. We are welcome to your participation. If you are interested in joining us, please do not hesitate contact us via e-mail.