Magnetic Resonance Imaging

MRI techniques are employed by investigators at Haskins Laboratories either to obtain data on the shape of the vocal tract during vowel, fricative and labial production or to identify the cerebral locus of speech production and perception activity by monitoring changes in the distribution of bloodflow in the brain. The equipment being used is a 1.5 Tesla SIGNA machine manufactured by the General Electric Company which is located in the Yale Medical School. The machine is capable of producing about 25 images of adjacent sections through the head in from 2 to 7 minutes of data acquisition time.

MRIs exploit the fact that, if a steady magnetic field is applied to a biological tissue, the magnetic moments of the protons within that tissue become aligned with the steady field. Then, if a pulse of radio frequency (rf) energy is injected at a specific frequency called the Larmor frequency, the protons are sent into resonance and their magnetic moment vectors precess around the steady field vector. The Larmor frequency is directly determined by the magnetic field strength B and is given by the expression f=[[gamma]]B/2[[pi]] where [[gamma]] is called the gyromagnetic ratio.

After receiving an excitation pulse, the magnetization of the tissue will have longitudinal and transverse components, longitudinal being along the z axis and transverse being in the xy plane. The longitudinal magnetization gradually recovers as energy is radiated and lost as heat. Meanwhile the transverse magnetization decays, a process that can be detected by an antenna. Four fundamental MR parameters are of importance in an MRI. They are the proton density [[rho]], the time constant T1 of the exponential recovery of longitudinal magnetization, the decay time constant T2 of the transverse magnetic moment vector and T2* the time constant of the free induction decay (FID) of transverse magnetization. These independent parameters are the properties of all biological tissues. Because they can differ in neighboring tissues, they can be used to distinguish one tissue from another. It is the relative contribution to the brightness of an MRI image that is made by these four parameters that determines how clearly one type of tissue can be seen in contrast to another.

The relative contributions of the four parameters are determined by the rf excitation and the magnetic environment in which the image data are collected. By the appropriate application of rf pulses and magnetic field gradients directed along the x and y axes, the signals emanating from small volume elements of tissue, termed voxels, can be coded in terms of the frequency and phase of the signals picked up by an antenna. A MRI is obtained by computing a 2DFT of a matrix consisting of 256 signals each sampled 256 times, each signal obtained with a different phase encoding. Each image can be a two-dimensional digital representation of a slice through a subject's vocal tract or brain. In the case of the vocal tract, the image is analyzed by superimposing it on a model of the vocal tract in the program CASY (see below). Alternatively, in the case of a brain image, it is examined for evidence of localized changes in the oxygenation state of hemoglobin (revealed by the decay time T2*) and signifying alterations in bloodflow and oxygen balance (Ogawa et al., 1990, 1992). Evidence of such changes indicate putative sources of the brain activity underlying articulation or language processing (Constable et al. 1993; McCarthy et al., 1993).

Back to Tools