{
"id": "mri-image-production",
"modality": "MRI",
"category": "Image Production: Sequences, Parameters and Artifacts",
"title": "MRI Image Production: Sequences, Parameters and Artifacts",
"minutes": 9,
"slides": [
{
"id": 1,
"diagram": "title",
"heading": "MRI Image Production",
"bullets": [
"106 of 200 scored questions (53%)",
"Physics 40, parameters 36, acquisition 30",
"Larmor, relaxation, weighting, k-space",
"SNR, scan time, sequences, artifacts"
],
"narration": "Welcome to MRI Image Production. This is the biggest category on the MRI boards: 106 of the 200 scored questions, or 53 percent. ARRT splits it into physical principles at 40 questions, sequence parameters and options at 36, and data acquisition, processing and storage at 30. We'll cover the Larmor equation, T1 and T2 relaxation, how T R and T E set weighting, and k-space. Then we'll work through signal to noise, scan time math, the main sequences, and the artifacts you have to recognize and fix. Let's get started.",
"svg": ""
},
{
"id": 2,
"diagram": "larmor",
"heading": "Larmor and resonance",
"bullets": [
"Frequency = gyromagnetic ratio x B0",
"Hydrogen: 42.58 MHz per tesla",
"63.86 MHz at 1.5 T, 127.7 MHz at 3 T",
"RF at Larmor tips M0; 90 degrees flips it transverse"
],
"narration": "Everything starts with the Larmor equation. Precessional frequency equals the gyromagnetic ratio times the main field strength, B zero. For hydrogen, the gyromagnetic ratio is 42.58 megahertz per tesla. So at 1.5 tesla, protons precess at 63.86 megahertz, and at 3 tesla, about 127.7 megahertz. Double the field, double the frequency. A classic trap is offering 63.86 as the 3 tesla answer. Resonance happens when the radio frequency pulse matches the Larmor frequency. It tips the net magnetization, M zero, and brings the protons into phase. A 90 degree pulse puts it all in the transverse plane.",
"svg": ""
},
{
"id": 3,
"diagram": "relaxation-curves",
"heading": "T1, T2 and T2*",
"bullets": [
"T1: Mz regrows to 63% (spin-lattice)",
"T2: Mxy decays to 37% (spin-spin)",
"T2* adds field inhomogeneity: always shortest",
"Fat short T1 (bright T1); fluid long T1 and T2"
],
"narration": "T1, or spin lattice relaxation, is the time for longitudinal magnetization to regrow to 63 percent of its maximum. T2, or spin spin relaxation, is the time for transverse magnetization to decay to 37 percent of its starting value. Don't swap those two numbers. T2 star adds the effect of field inhomogeneity and susceptibility, so T2 star is always shorter than T2, and T2 is shorter than or equal to T1. Fat has a short T1, so it is bright on T1. Water and cerebrospinal fluid have a long T1 and a long T2: dark on T1, bright on T2.",
"svg": ""
},
{
"id": 4,
"diagram": "tr-te-grid",
"heading": "TR, TE and weighting",
"bullets": [
"TR controls T1; TE controls T2",
"T1: short TR (400-700 ms), short TE (10-20 ms)",
"T2: long TR (2000+ ms), long TE (80-120 ms)",
"PD: long TR, short TE",
"GRE: big flip + short TE = T1; small flip = T2*"
],
"narration": "Repetition time, or T R, controls T1 weighting. Echo time, or T E, controls T2 weighting. A spin echo T1 uses a short T R, about 400 to 700 milliseconds, and a short T E, about 10 to 20 milliseconds. A T2 uses a long T R, 2,000 milliseconds or more, and a long T E, about 80 to 120 milliseconds. Proton density uses a long T R with a short T E. In gradient echo, flip angle acts like T R: a large flip angle with a short T E gives T1, and a small flip angle with a longer T E gives T2 star.",
"svg": ""
},
{
"id": 5,
"diagram": "k-space",
"heading": "Spatial encoding and k-space",
"bullets": [
"Slice select: gradient on during the RF pulse",
"Thin slice: steeper gradient or narrower transmit BW",
"Phase encode before readout; frequency during echo",
"k-space center = contrast; edges = detail",
"Phase steps set scan time, not frequency matrix"
],
"narration": "Spatial encoding uses three gradients. Slice selection turns a gradient on during the radio frequency pulse, so only one slab is excited. For a thinner slice, use a steeper slice select gradient or a narrower transmit bandwidth. Phase encoding is applied after excitation and before readout. Frequency encoding happens while the echo is collected. All of that raw data lands in k-space. The center of k-space sets signal and contrast, and the edges set fine detail. And remember, the number of phase encoding steps sets two D scan time. The frequency matrix does not.",
"svg": ""
},
{
"id": 6,
"diagram": "snr-time",
"heading": "SNR, resolution and time",
"bullets": [
"SE time = TR x phase steps x NEX",
"FSE divides by ETL; 3D multiplies by partitions",
"Double NEX: SNR up 1.41 times, time doubles",
"Halve receive bandwidth: SNR up 1.41 times",
"Bigger voxel = more SNR, less resolution"
],
"narration": "Now the math. Spin echo scan time equals T R times phase encoding steps times NEX, the number of excitations. Fast spin echo divides that by the echo train length, and 3D multiplies it by the number of slice partitions. Signal to noise follows the square root of NEX, so doubling NEX raises SNR only about 1.41 times while doubling scan time. Halving the receive bandwidth also raises SNR about 1.41 times, but it increases chemical shift. Finally, a bigger voxel, from a larger field of view, a smaller matrix or a thicker slice, means more SNR but less spatial resolution.",
"svg": ""
},
{
"id": 7,
"diagram": "se-vs-gre",
"heading": "Core pulse sequences",
"bullets": [
"Spin echo: 180 refocuses, true T2, not T2*",
"FSE: echo train fills many lines per TR",
"GRE: gradient echo, low flip, T2* sensitive",
"EPI: fastest, most susceptibility distortion",
"DWI: restriction bright on DWI, dark on ADC"
],
"narration": "Spin echo uses a 90 degree pulse, then a 180 degree refocusing pulse that cancels fixed inhomogeneity, so the contrast is true T2, not T2 star. Fast spin echo, or turbo spin echo, uses a train of 180s, filling several phase lines per T R. Gradient echo forms its echo with a gradient reversal instead of a 180, uses low flip angles, and is sensitive to T2 star. Echo planar imaging is the fastest sequence but has the most susceptibility distortion. It drives diffusion imaging, where restricted water is bright on D W I and dark on the apparent diffusion coefficient, or A D C, map.",
"svg": ""
},
{
"id": 8,
"diagram": "inversion-recovery",
"heading": "Inversion and fat suppression",
"bullets": [
"Tissue nulls at TI = 0.69 x its T1",
"STIR nulls fat: TI about 150-170 ms at 1.5 T",
"Never STIR after gadolinium",
"FLAIR nulls CSF: TI about 2000-2500 ms",
"Spectral fat sat fails near metal; use STIR/Dixon"
],
"narration": "Inversion recovery starts with a 180 degree inversion pulse. After the inversion time, T I, a 90 degree pulse reads the signal, and a tissue is nulled when T I is about 0.69 times its T1. STIR nulls fat with a T I of about 150 to 170 milliseconds at 1.5 tesla. Because STIR nulls anything with a short T1, never use it after gadolinium, or you'll erase the enhancement. FLAIR nulls cerebrospinal fluid with a T I around 2,000 to 2,500 milliseconds. Spectral fat sat needs a homogeneous field, so it fails near metal. Use STIR or Dixon instead.",
"svg": ""
},
{
"id": 9,
"diagram": "artifact-table",
"heading": "Artifacts and fixes",
"bullets": [
"Aliasing: phase direction; enlarge FOV, oversample",
"Motion and flow ghosts: always phase direction",
"Chemical shift: frequency; wider BW reduces it",
"Susceptibility: worst on GRE/EPI; use FSE near metal",
"Zipper: RF leak, check the door and seal"
],
"narration": "Artifacts are heavily tested. Aliasing, or wrap, folds anatomy outside the field of view onto the opposite side in the phase direction. Fix it with a larger field of view or phase oversampling. Motion and flow ghosts always propagate in the phase direction. Chemical shift sits in the frequency direction, because fat and water are about 3.5 parts per million apart. A wider receive bandwidth reduces it, and a narrower one makes it worse. Susceptibility artifact is worst on gradient echo and echo planar imaging, so use fast spin echo near metal. A zipper line means radio frequency leakage: check the door and its seal.",
"svg": ""
},
{
"id": 10,
"diagram": "summary",
"heading": "Lock these in",
"bullets": [
"Larmor: 63.86 MHz at 1.5 T, 127.7 MHz at 3 T",
"TR sets T1, TE sets T2; T1 63%, T2 37%",
"Scan time = TR x phase steps x NEX (FSE / ETL)",
"Double NEX = 1.41 times SNR, twice the time",
"Ghosts and wrap: phase; chemical shift: frequency"
],
"narration": "Let's lock it in. The Larmor frequency is 63.86 megahertz at 1.5 tesla and 127.7 megahertz at 3 tesla. T R sets T1 weighting and T E sets T2 weighting, with T1 at 63 percent recovery and T2 at 37 percent decay. Scan time equals T R times phase steps times NEX, divided by the echo train length for fast spin echo. Doubling NEX gives about 1.41 times the signal to noise for twice the time. And motion ghosts and wrap run in the phase direction, while chemical shift runs in the frequency direction. Nice work.",
"svg": ""
}
]
}