Jong-Hwan Lee, PhD
Instructor in Radiology
Brigham and Women’s Hospital
Harvard Medical School, Boston, MA
Mentor Dr. Yoo
My recent research has focused on the application of the ANN and machine learning algorithms to the biomedical signals, such as functional MRI and electroencephalography (EEG). I was introduced to the field of biomedical research during my post-doctoral work under the guidance of Dr. Seung-Schik Yoo in the Department of Radiology, Brigham and Women’s Hospital (BWH). My work as a research fellow [Research funding] on Dr. Yoo’s NIH-funded research project has drawn on my extensive experiences on the development of signal processing techniques and computer programming skills. I have been involved in the acquisition, analysis of fMRI data, development of user-friendly and comprehensive software toolboxes, and manuscript preparation for this project. I have also been engaged in a research study on hormones and brain activity in women with and without depression (P.I. Dr. Jill Goldstein in the Division of Women’s Health, BWH). In Dr. Goldstein’s project, I am also participating in data acquisition, analyses, and giving technical expertise regarding integration of psychophysiological data (skin conductance & heart rate) with fMRI data.
A major outcome of the project with Dr. Yoo was a recent publication of our real-time automated registration technique and concurrent monitoring method of fMRI data [article #9]. Using this technique, the feasibility of the fMRI-mediated learning and consolidation of brain activations was shown within a primary motor area [article #7 and #12] and primary auditory area [article #8] via real-time fMRI (rtfMRI) neurofeedback modality.
In another recent publication [article #10 and #11], I introduced a novel group fMRI analysis technique using an independent vector analysis (IVA) method. The IVA algorithm is more advanced than the ICA algorithm since IVA allows an additional dependent parameter for the analysis compared to ICA, and thus this parameter could be assigned to the index of a subject from group fMRI data. While keeping the advantageous non-parametric and multivariate characteristics of the ICA method, the IVA method proposed presented an additional merit of automatic grouping of similar activation patterns among the subjects, which can be beneficially utilized in the group inference of fMRI data. Moreover, the IVA method showed better accuracy on the estimation of hemodynamic responses and subsequent improvement of the statistical significance compared to other conventional methods. The developed IVA method may be particularly effective for the analyses of subject- and region-specific abnormal BOLD signals such as those due to neuroleptic medications and substance abuse.
In future research activities, I will concentrate more on pre-clinical applications based on the techniques developed. These applications include (1) the motor rehabilitation of stoke patients, (2) feasibility study on the treatment of substance abuse using rtfMRI neurofeedback, (3) real-time monitoring and analysis of dynamic contrast-enhanced breast MRI for early detection of malignant tumors (in collaboration with Dr. Eva Gombos at the Lee Bell Center for Breast Imaging, BWH), and (4) identification of cortico-striatum neural circuitries of the schizophrenic patients using the developed IVA technique (in collaboration with Dr. Cynthia Wible at the Department of Psychiatry, HMS). Based on these research collaborations, I am very interested in the development of comprehensive and user-friendly software toolboxes and dissemination of these applications to the research community.