Mo Shahdloo

I am a computer vision and machine learning scientist at Vicon. Before that, I was a researcher at Oxford Centre for Functional MR Imaging of the Brain (FMRIB), University of Oxford, mentored by Mark Chiew and Karla Miller.

Going back, I was trained in Electrical Engineering at Amirkabir University and Sharif University, Tehran. After working on developing industrial embedded systems for oil & gas plants for couple of years, I moved to Ankara where I did my graduate work in the lab of Tolga Çukur, working on MRI and computational neuroscience.

My current research centers on developing machine learning tools for computer vision. Stretching from the past, I am still interested in developing methods to make MR imaging faster and more accurate, which involve physics-based as well as deep-learning based approaches. Recently, I was working on reducing motion artifacts and thermal noise in functional MRI in awake non-human primates and humans. I am also interested in building computational models to map representation of visual and auditory natural stimuli in the human brain, using Bayesian statistics and deep learning.
Below, you can peek into the stuff I have been busy doing recently.

Research

  • Motion-correction for awake non-human primate fMRI
    Functional imaging of awake non-human primates is challenging not only due to their smaller brain size, but also due to excessive local field fluctuations due to animals' motion. Using a novel pulse sequence, efficient reconstruction techniques, and a custom-made multi-channel coil we aim to perform accelerated functional scans of awake monkeys.

  • Attentional modulations of action-category representation in the brain
    We used voxelwise models to map the semantic representation of hundreds of objects and actions across cortex. We then studied the attentional modulation of the semantic representation during action-based visual search.

  • Temporal receptive windows in the brain mapped via deep language networks
    Using a natural story listening experimental paradigm, we use computational models based on deep neural networks to study the effect of context integration on the representation of a large dictionary of words across the cortex. We are further interested in studying the attentional modulations of this property across cortex.

  • Rapid self-tuning compressed sensing MRI
    While parameter selection is critical for compressed-sensing reconstruction quality, the optimal parameters are subject and dataset specific. Thus, commonly practiced heuristic parameter selection generalizes poorly to independent datasets. We proposed a new self-tuning compressed-sensing reconstruction method that uses computationally efficient projections onto epigraph sets of the l1 and total-variation norms to simultaneously achieve parameter selection and regularization.

  • Trade off between fat-suppression and partial-voluming in bSSFP acquisitions
    Fat-suppression in bSSFP can be achieved using weighted average of in-phase and out-of-phase acquisitions raised to a negative power. We studied the effects of the power parameter on the trade off between fat-suppression and partial voluming under various experimental setups. We devised optimal parameter values meeting the fat-suppression and partial voluming requirements.

Teaching

  • K-space and image reconstruction
    This talk provides some intuition about the concept of spatial encoding in MRI by introducing the k-space, and elaborates more on some of the special sampling scenarios. Also, introduction to baiscs of parallel imaging is provided. This talk was presented during the educational course at ISMRM British Chapter MR-Fest 2021



Publications

Please use this bibliography file to cite any of the works below.

Articles

Conference publications

Open Reviews

To encourage transparency of the academic publication ecosystem, I sign the manuscript peer reviews that I am solicited to do, and publish them both here and on PubPeer whenever possible (i.e., when a preprint is posted online).

CV

Here is my recent CV -> in HTML , and in PDF

Without music, life would be a mistake ― Friedrich Nietzsche