Practical Magnetic Resonance Imaging II

Course Name: 
G16.4428 - Practical Magnetic Resonance Imaging II
Course Director/Instructors: 

Ricardo Otazo, PhD
Henry Rusinek, PhD
Jelle Veraart, PhD
James Babb, PhD

Course Outline: 

This course is a practical introduction to image reconstruction, processing, and analysis in magnetic resonance imaging (MRI). The course is divided in three modules. The first part focuses on MR image reconstruction, with in-depth mathematical descriptions of the most common algorithms. The second part introduces students to the basics of medical image representation and analysis. This includes post-processing algorithms that are highly relevant in clinical radiology: tissue/organ segmentation, coregistration and kinetic modeling of dynamic MRI. The third part provides an introduction to selected topics in biostatistical concepts and reasoning. During laboratory sessions and homework, students will use Matlab and FireVoxel to implement and test image reconstruction methods, perform image segmentation and coregistration. Prerequisites are G16.4427 or permission of the course director for students not enrolled in the Sackler training program in biomedical imaging.

  • Teach students basic and advanced methods for MR image reconstruction
  • Present clinically relevant concepts of medical image processing
  • Provide an introduction to core applied statistical concepts and methods 
  • Review the mathematics behind the most common techniques and expose students to practical examples 


  • Introduction to the course
  • Spatial encoding and k-space
  • Sampling theorem and aliasing
  • Fourier image reconstruction
  • Point spread function and image resolution
  • Zero-padding and windowing
  Download Lecture 1  
  • Review of Matlab commands
  • Generate k-space representation of an image and recover original image
  • Effects of under sampling, zero-padding and windowing
  • Conjugate symmetry
  • Margosian and homodyne reconstruction
  • Iterative partial Fourier reconstruction
  Download Lecture 2  
  • Margosian reconstruction
  • POCS algorithm
  • Radial and spiral sampling trajectories
  • Regridding and density compensation
  • Backprojection reconstruction
  Download Lecture 3  
  • Regridding reconstruction of radial and spiral k-space data
  • Non-uniform FFT
  • Multicoil systems and coil sensitivity encoding
  • SENSE parallel imaging reconstruction
  • g-factor
  Download Lecture 4  
  • Estimation of coil sensitivities
  • SENSE reconstruction of an under sampled brain image
  • SMASH parallel imaging reconstruction
  • GRAPPA parallel imaging reconstruction
  Download Lecture 5  
  • GRAPPA kernel reconstruction
  • GRAPPA reconstruction of an undersampled brain image
  • Iterative partial Fourier reconstruction
  • Compressed sensing theory
  • MRI compressibility and sparsifying transforms
  • Incoherent k-space sampling
  • Non-linear reconstruction: iterative soft-thresholding
  • Combination of compressed sensing and parallel imaging
  Download Lecture 6  
  • Random k-space under sampling of MRI data
  • Reconstruction using iterative soft-thresholding
  • Spatiotemporal correlations and sparsity
  • Temporal parallel imaging: TSENSE and TGRAPPA
  • k-t SENSE and k-t GRAPPA
  • Compressed sensing for dynamic imaging
  Download Lecture 7  
  • MR signal models: B0 inhomogeneity, T1 and T2 reconstruction
  • Iterative model-based reconstruction algorithms
  Download Lecture 8  
  • X-ray, CT, ultrasound, gamma cameras, PET/SPECT
  • Ionizing radiation, medical radiation exposure
  • Image representation, file formats, DICOM, Analyze formats
  • Look-up tables, organizing slices into volumes
  • regions of interest
  Download Lecture 9  
  • understanding DICOM files Analyze format
  • Image histogram, computations of histogram features
  • assessing SNR
  • Sources of image nonuniformity
  • Smooth multiplicative model of the bias field
  • Homeopathic filtering approach
  • Tissue intensity model
  • Histogram-based model, N3 method
  • Validation of non uniformity algorithms
  Download Lecture 10  
  • Spatial filtering and convolutions
  • Spatial transformation (translations/rotations/affine)
  • ROIs and morphological manipulations
  • The role of segmentation in medical imaging
  • Manual ROI editing
  • Thresholding
  • Erosion and dilation operators
  • Region growing
  Download Lecture 11  
  • Parameters affecting image noise
  • Basic concepts of probability distributions: expectation and variance
  • Gaussian and Rayleigh distributions, simulating pseudorandom noise in Real, Imaginary, and Magnitude images
  • Rician distribution
  • Implication for parametric mapping (T2-mapping, ADC)
  • Denoising filters
  • Physiologic vs thermal noise
  • Noise in parallel imaging
  Download Lecture 11  
  • FireVoxel segmentation tools
  • Improving segmentation by filtering and uniformity correction
  • White matter lesion load: multi-stage segmentation
  • Validation of segmentation methods: Hausdorf distance and Dice coefficients
  • Estimating T1, T2, and ADC
  • Non-linear fitting for IVIM models
  • DCE-MRI models with arterial input for tumor perfusion
  Download Lecture 13  
  • Classification of registration problems
  • Landmark methods
  • Cost functions
  • Optimization strategies
  • Mutual Information
  • Validation of coregistration methods
  Download Lecture 14  
  • Landmark matching
  • Correlational approach
  • Mutual Information applied to dynamic imaging
  • Henry Rusinek, PhD and Jelle Veraart, PhD
  • Accuracy versus precision
  • The Gaussian probability distribution
  • Type I and Type II errors
  • Rayleigh distribution
  • Rician distribution
  • James Babb, PhD
  • P-value
  • Power and sample size in study design
  • Analysis of variance (ANOVA) and t-test
  • Regression and correlation
  • Non-parametric statistical methods
  • James Babb, PhD
  • Interpret differences in data distributions via visual display
  • Calculate standard scores and resulting probabilities
  • Calculate and interpret confidence intervals for population means
  • Perform a two-sample t-test and interpret the results
  • Perform an ANOVA and interpret the results
  • James Babb, PhD

The course meets twice per week and it is organized as fifteen 120-minute lectures, eleven 180-minute labs, and two exams. Students will be evaluated based on course participation (10%), two exams (20% each), and laboratory projects (50%).


Latest Updates

Philanthropic Support

We gratefully acknowledge generous support for radiology research at NYU Langone Health from:
• The Big George Foundation
• Raymond and Beverly Sackler
• Bernard and Irene Schwartz

Go to top