Qolamreza 'Ray' Razlighi, Ph.D.
Assistant Professor in Neurology
Adjunct Assistant Professor in Biomedical Engineering
Columbia University Medical Center
630 West 168th Street, P&S Box 16
New York, NY 10032
- Pre-processing of BOLD fMRI data. The problem of fMRI pre-processing is profound and urgently needs rigorous attention to ensure the continuing high quality of basic and clinical neuroimaging research in the future, particularly given the propensity for ever-more complex derivation of functional-connectivity based biomarkers using resting-state fMRI. Most of the current pre-processing algorithms have originated from task-based fMRI data analysis and have subsequently been adopted for resting-state data analysis as well. However, a fundamental complication was overlooked at first: task-based fMRI data analysis is usually performed by correlating the fMRI data with an externally given task time series. Such an external measure of activation usually does not exist for functional-connectivity studies using resting-state data; instead, intrinsic “seed” measures such as a voxel signal, or the mean signal of a set of voxels, are considered. Since both seed signals and remaining brain signals are affected by the same artifactual processes during data acquisition, there is a danger of false-positive inflation of spurious functional-connectivity findings at rest, as has been shown in many recent studies that have focused on the effects of motion, respiration, etc.
- Region-based spatial normalization. A major longstanding problem in functional neuroimaging studies of cognitive aging is that the large age-related changes in brain morphology make it difficult to co-register brains, a key step for studies comparing task-related activation in young and old groups. To address this issue, I propose to develop a region-based spatial normalization (RBSN) technique that will increase the accuracy of the fMRI data localization. RBSN aligns each neuroanatomical region of the human brain separately. In contrast, the prevailing spatial normalization method tries to align all regions of the brain at once. RBSN will therefore provide more accurate localization of activation and ensure that group analyses test the same brain area in each study participant. Better between-participant registration also provides additional statistical power to detect activation in regions that may not have reached the significance level using prevailing methods (i.e. reduce type II error). It may also rule out previously noted areas of activation that were detected for artifactual reasons (i.e., reduce type I error).
- Multilateral Markov Random Field and Brain Subcortical Segmentation. Conventional Markov random field is hampered by noncausality and its causal definitions are also not free of difficulties. For instance, the Markov mesh random field has strong diagonal dependency in its local neighboring system. I have introduced multilateral Markov random field (MMRF) to overcome this issue. MMRF is a MRF whereas the reverse may not hold. The joint distribution of a causal MRF is readily given by the product of the low-dimensional local distribution whereas in conventional MRF it is only given through Gibbs distribution. Low-dimensional joint pdf's are often estimated using a joint histogram for homogeneous field or by a few sample of the field for inhomogeneous fields. This makes the model closely tied to the image in use. So far MMRF is used for the computation of image entropy and mutual information.
1. Q. R. Razlighi, N. Kehtarnavaz, and A. Nosratinia, “Computation of image spatial entropy using quadrilateral Markov random field”, IEEE Transaction on Image Processing, vol. 18, no. 12, Dec. 2009
2. Q. R. Razlighi, A. Orekhov, A. Laine, and Y. Stern, “Causal MRF for Brain MR Image Segmentation,” Proceedings of IEEE Engineering in Medicine and Biology Society, San Diego, Aug. 2012.
3. Q. R. Razlighi, J. Steffener, C. Habeck, A. Laine, and Y. Stern, “Resting State Inter and Intra Hemispheric Human Brain Functional Connectivity,” Proceedings of IEEE Engineering in Medicine and Biology Society, Osaka, Japan, Jul. 2013.
4. Q. R. Razlighi and N. Kehtarnavaz, “Spatial Mutual Information as Similarity Measure for 3D Brain Image Registration”, IEEE Journal of Translational Engineering in Health and Medicine, vol.2, no., pp.27-34, 2014.
5. Q. R. Razlighi, C. Habeck, J. Steffener, Y. Gazes, L. B. Zahodne, A. M. Brandt, and Y. Stern, “Unilateral disruptions in the default network with aging in native space,” Brain and Behavior , vol. 4, no. 2, pp. 143-157, 2014.