ANDERSON Timothy

PhD Student

at AMU

PHENIQS team

PhD Student

Detailed Activities

My PhD focuses on reducing the inhomogeneities of the perturbative RF magnetic field (“B_1”) in the MRI signal in human brains when using the inhomogeneous Magnetization Transfer (ihMT) contrast at 7 T (≡kg⋅s^(-2)⋅A^(-1), a tesla is the unit of magnetic flux density).
While the B_1 field has global homogeneity within the brain at low density, wave-like interaction effects such as constructive and destructive interferences prop up as the B_1 wavelength λ_1 reaches the same scale as the characteristic length of the human brain λ_brain=15 cm. This transition disrupts the global homogeneity of the field which becomes locally homogeneous only.
This would not be so problematic if the relationship between the B_1 field and the MRI ihMT signal was a linear relationship (ihMT_signal= f(B_1,…)∝〖a×B〗_1+c,with (a,b)∈R^2). However, due to their complicated relationship, we have yet to find an analytical relationship between the two. Our current models only work at specific scales (1.5T, 3T, 7T…) by making rough and possibly unwarranted assumptions about the effect of B_1 on the MRI signal.

The goal is manifold:
1. Find an equation that models the relationship between B_1 and the ihMT signal at 7 T
2. Extend the model to the largest density interval possible (e.g., 1 T to 15 T)
3. Compensate the ihMT signal for the B_1 inhomogeneities using the model
4. Extract quantitative parameters that best characterize MRI images from the model
5. And what other possible avenues have we yet to foresee?

Finding a tractable analytical equation that models the role of B_1 in the ihMT MRI signal may be impossible. Instead, we may need to rely on semi-analytical formalisms to get us there. These formalisms may include fitting generic functions to empirical data, computer simulations, arcane mathematical techniques such as Effective Field Theories (EFT), or a mix of everything.
It is my hope that the Math tools and CS skills I learned from working in particle Physics (esp. flavor violating decays of charged leptons in the SMEFT) and data science will be of help in finding a generalized model of the ihMT MRI signal.

Keywords

  • Inhomogeneous Magnetization Transfer (ihMT)
  • 7T human MRI
  • Artificial Intelligence, High Performance Computing (HPC)
  • Computational Physics, Mathematical Physics
  • Flavor Physics, Effective Field Theories (EFT)

Teaching

EPF – School of Engineering: Data Science major (graduate courses)
• Distributed Computing
• Programming and Agile Development
• Introduction to Machine Learning
• Predictive Modeling
• Unsupervised Clustering

Research Projects

Publications :