subscriber@gdayradiographer.com.au • subscriber@gdayradiographer.com.au • subscriber@gdayradiographer.com.au • subscriber@gdayradiographer.com.au

Lesson 1.5: Free Induction Decay (FID) and K-space Fundamentals

Learning Objectives

Upon completion of this lesson, radiographers will be able to:

Analyse the physics underlying Free Induction Decay (FID) signal generation and detection
Evaluate the relationship between transverse magnetisation decay and signal characteristics
Critically assess K-space architecture and its impact on image quality parameters
Apply Fourier transform principles to MR image reconstruction processes
Optimise acquisition parameters based on K-space filling strategies

Introduction

The transition from RF excitation to image formation represents a fundamental concept in MRI physics. The Free Induction Decay (FID) signal emerges immediately following RF pulse termination, encoding spatial information through frequency and phase modulation. This raw electromagnetic signal, captured in the spatial frequency domain known as K-space, undergoes mathematical transformation to produce diagnostic images. Mastery of these concepts is essential for advanced MRI practice and forms the foundation for understanding pulse sequence design, artefact recognition, and protocol optimisation.

💡 Plain English: What is FID?

Imagine striking a bell. The moment you stop hitting it, the bell continues to ring and gradually fades away. That's exactly what happens with hydrogen protons after an RF pulse!

The RF pulse is like striking the bell (exciting the protons)

The protons "ring" at their Larmor frequency as they precess together

The signal fades away as protons lose synchronisation (dephasing)

We capture this "ringing" in our receiver coil - that's the FID signal!

1. Free Induction Decay: Physics and Signal Characteristics

1.1 Definition and Physical Basis

Free Induction Decay represents the time-dependent electromagnetic signal generated by precessing transverse magnetisation (M_xy) following RF excitation:

Free: Signal evolves without external RF influence
Induction: Electromagnetic induction in receiver coil (Faraday's Law)
Decay: Exponential signal reduction due to dephasing mechanisms

1.2 Signal Generation Mechanism

The FID signal arises through the following sequence:

RF excitation creates phase-coherent transverse magnetisation
RF termination allows free precession at Larmor frequency
Dephasing processes cause progressive signal decay:
- T2 relaxation: Irreversible spin-spin interactions
- T2 effects*: Additional dephasing from B₀ inhomogeneities
- Chemical shift: Frequency differences between molecular environments
Electromagnetic induction in receiver coil produces measurable voltage

💡 Plain English: T2 vs T2 - What's the Difference?*

Think of a marching band trying to stay in step:

T2 decay = Band members naturally falling out of step due to random bumping into each other (intrinsic, unavoidable)

T2 decay* = Same thing PLUS the field being uneven, so some marchers have to walk uphill (external magnetic field imperfections)

T2 is always shorter than T2* because it includes both effects!
T2* = T2 + Field Inhomogeneity Effects

1/T2* = 1/T2 + 1/T2'
        ↑       ↑
   intrinsic  field
   (tissue)   imperfections

1.3 Mathematical Representation

<details> <summary>📐 Simple Version: Signal = Starting strength × Decay factor × Oscillation</summary>

Full Mathematical Form:

The FID signal can be expressed as:

S(t) = S₀ · e^(-t/T2*) · e^(iωt)

Where:

S₀ = Initial signal amplitude (how strong the signal starts)
T2* = Effective transverse relaxation time (how fast it decays)
ω = Larmor frequency (how fast protons spin)
t = Time after RF pulse
e^(iωt) = Complex oscillation term (encodes frequency information)

Breaking it down:

e^(-t/T2*) → Exponential decay envelope (the "fading" part)
e^(iωt) → Sinusoidal oscillation (the "ringing" part)

</details>

📊 Visual: FID Decay Curve

Signal
Amplitude
    │
S₀  │█
    │██
    │███
    │█████
    │███████
    │██████████
    │██████████████
    │██████████████████████
    │█████████████████████████████████████
    │████████████████████████████████████████████████████████████
    └──────────────────────────────────────────────────────────────▶ Time
         T2*        2×T2*       3×T2*       4×T2*      5×T2*
              ↑
         Signal drops to 37% of original (1/e)

    At T2*:   37% signal remaining
    At 2×T2*: 14% signal remaining
    At 3×T2*:  5% signal remaining
    At 5×T2*: <1% signal remaining (essentially zero)

2. Signal Detection and Digitisation

2.1 Receiver Coil Configuration

Quadrature detection: Two orthogonal coils detect M_x and M_y components
Bandwidth optimisation: Receiver bandwidth must encompass all frequencies within FOV
Signal-to-noise considerations:
- SNR ∝ 1/√(Bandwidth)
- Trade-off between coverage and signal quality

2.2 Analog-to-Digital Conversion

Critical parameters for signal digitisation:

Sampling rate: Must satisfy Nyquist criterion (≥2× highest frequency)
Dynamic range: 12-16 bit ADC typical for clinical systems
Dwell time: Time between successive sampling points
- Dwell time = 1/(2 × Receiver Bandwidth)

💡 Plain English: The Nyquist Criterion

Imagine trying to photograph a spinning fan:

Take photos too slowly → Fan blades look like they're going backwards or standing still (aliasing!)

Take photos fast enough → You can see the true motion

The Rule: Sample at least 2× faster than your highest frequency
Example:
If your highest frequency is 32 kHz:
Minimum sampling rate = 2 × 32 kHz = 64 kHz

If you sample slower → ALIASING ARTEFACT
(Signal "wraps around" and appears in wrong location)

⚠️ Clinical Takeaway: Bandwidth Trade-offs

Wider Bandwidth Narrower Bandwidth
✅ Less chemical shift artefact ❌ More chemical shift artefact
✅ Shorter TE possible ❌ Longer minimum TE
❌ Lower SNR ✅ Higher SNR
✅ Less distortion ❌ More susceptibility

Clinical rule: Use wider bandwidth for areas with metal/air interfaces (sinuses, spine hardware). Use narrower bandwidth when SNR is critical (small FOV, high resolution).

3. K-space: The Spatial Frequency Domain

3.1 Conceptual Framework

K-space represents a mathematical construct where:

Each point contains amplitude and phase information
Position determines spatial frequency content
Data acquisition fills K-space systematically

💡 Plain English: K-space as a Recipe Book

Think of K-space as a recipe book for building your MRI image:

Centre of K-space = Basic ingredients (flour, sugar, eggs)

Gives you the overall shape and contrast of the image

Without it, you have nothing recognisable

Edges of K-space = Decorating instructions (icing details, sprinkles)

Gives you sharp edges and fine details

Without it, image looks blurry but still recognisable

Key insight: You could bake a cake with ONLY basic ingredients (centre K-space) and it would still BE a cake. But you can't make a cake from ONLY decorations (edge K-space)!

This is why central K-space is so important for image quality.

3.2 K-space Architecture

Central K-space (Low spatial frequencies)

Contains ~90% of image signal intensity
Determines tissue contrast (T1, T2, PD weighting)
Affects SNR and contrast-to-noise ratio (CNR)
Sampling density impacts image brightness

Peripheral K-space (High spatial frequencies)

Contains edge information and fine detail
Determines spatial resolution
Affects image sharpness and edge definition
Under-sampling causes blurring or truncation artefacts

📊 Visual: K-space Centre vs Periphery

                    K-SPACE ANATOMY

            High ky (top edges)
                    ↑
                    │
    ┌───────────────────────────────────┐
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│ ← Fine vertical details
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    │░░░░░░░░░░┌─────────────┐░░░░░░░░░░│
    │░░░░░░░░░░│█████████████│░░░░░░░░░░│
    │░░░░░░░░░░│█████████████│░░░░░░░░░░│
←───│░░░░░░░░░░│██ CENTRE ███│░░░░░░░░░░│───→
Low │░░░░░░░░░░│█████████████│░░░░░░░░░░│   High kx
kx  │░░░░░░░░░░│█████████████│░░░░░░░░░░│   (right edges)
    │░░░░░░░░░░└─────────────┘░░░░░░░░░░│
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    │░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░│
    └───────────────────────────────────┘
                    │
                    ↓
            Low ky (bottom edges)

    ████ = CENTRE: Contrast, brightness, SNR (90% of signal!)
    ░░░░ = PERIPHERY: Sharp edges, fine details, resolution

3.3 K-space Coordinates

<details> <summary>📐 Simple Version: Gradients control WHERE we sample in K-space</summary>

Full Mathematical Representation:

k_x = γ ∫G_x(t)dt   (frequency encoding direction)
k_y = γ ∫G_y(t)dt   (phase encoding direction)
k_z = γ ∫G_z(t)dt   (slice selection or 3D encoding)

Where:

γ = gyromagnetic ratio (42.58 MHz/T for hydrogen)
G = gradient amplitude (mT/m)
∫dt = integral over time (area under gradient curve)

What this means practically:

Stronger gradient → Move faster through K-space
Longer gradient → Travel further in K-space
Positive gradient → Move in positive k direction
Negative gradient → Move in negative k direction

The gradient "steers" us through K-space like a joystick!

</details>

4.1 Frequency Encoding (Readout Gradient)

Applied during signal readout
Creates linear frequency variation across FOV
Determines K-space traversal along k_x axis
Gradient amplitude affects:
- FOV_x = 2π/(γ·G_x·Δt)
- Resolution = FOV/Matrix size

4.2 Phase Encoding

Applied between excitation and readout
Imparts phase shift proportional to position
Incremented stepwise across acquisitions
Number of phase steps = Phase matrix size
Phase encoding steps determine:
- Scan time ∝ N_phase × TR × NEX
- Phase direction resolution

💡 Plain English: Frequency vs Phase Encoding

Frequency encoding (readout): Happens during ONE echo

Like taking a photo with a panoramic sweep

All positions along this direction captured simultaneously

FAST - done in milliseconds

Phase encoding: Requires MULTIPLE repetitions

Like taking separate photos, each with a different "label"

Must repeat the sequence for each line in this direction

SLOW - determines total scan time
Why motion affects phase encoding more:

Frequency encoding: ----[ALL DATA IN ONE ECHO]----
                             ↓
                    Motion blurs within ONE instant
                    (usually minimal)

Phase encoding:     [Line 1]...[Line 2]...[Line 3]...
                      ↓          ↓          ↓
                    If patient moves between lines,
                    each line shows different position!
                    → GHOSTING ARTEFACT

4.3 K-space Trajectories

Cartesian Sampling

Line-by-line acquisition
Standard for most clinical protocols
Advantages: Motion insensitivity, artefact predictability
Disadvantages: Longer acquisition times

Non-Cartesian Sampling

Radial: Spokes from centre outward
Spiral: Continuous curved trajectory
Advantages: Motion robustness, efficient coverage
Disadvantages: Complex reconstruction, gradient demands

📊 Visual: K-space Filling Patterns

    CARTESIAN               RADIAL                  SPIRAL
    (Line-by-line)          (Spokes)               (Continuous)

┌─────────────────┐   ┌─────────────────┐   ┌─────────────────┐
│─────────────────│   │        │        │   │    ╭──────╮     │
│                 │   │   ╲    │    ╱   │   │  ╭─╯      ╰─╮   │
│─────────────────│   │    ╲   │   ╱    │   │ ╭╯    ●─────╯   │
│                 │   │     ╲  │  ╱     │   │ │   ╭───╯       │
│─────────────────│   │──────╲─┼─╱──────│   │ │ ╭─╯           │
│                 │   │       ╲│╱       │   │ │╭╯             │
│─────────────────│   │        ●        │   │ ●              │
│                 │   │       ╱│╲       │   │                 │
│─────────────────│   │      ╱ │ ╲      │   │                 │
└─────────────────┘   └─────────────────┘   └─────────────────┘

Each line = 1 TR        Each spoke = 1 TR    Continuous acquisition
Total time:             Total time:          Total time:
 N_phase × TR            N_spokes × TR        Single or few TRs

PROS:                   PROS:                 PROS:
• Simple                • Oversamples centre  • Very fast
• Predictable           • Motion robust       • Oversamples centre
• Robust                • Short TE possible   • Motion robust

CONS:                   CONS:                 CONS:
• Slow                  • Streaking artefacts • Off-resonance blur
• Motion sensitive      • Complex recon       • Gradient demanding

5. Fourier Transform: From K-space to Image Space

5.1 Mathematical Principles

<details> <summary>📐 Simple Version: Fourier Transform = "Unmixing" frequencies into an image</summary>

The Big Picture:

Think of K-space data as a symphony recording - all instruments playing together. The Fourier Transform is like separating that into individual instruments and placing each in its correct seat in the orchestra!

Full Mathematical Form (2D Fourier Transform):

Image(x,y) = ∫∫ K-space(k_x,k_y) · e^(i2π(k_x·x + k_y·y)) dk_x dk_y

What it does:

Takes EVERY point in K-space
Multiplies by a "detector wave" for each image location
Adds them all up
Result = brightness at that image location

Key properties:

Linearity: Signal averaging works (add K-spaces → averaged image)
Shift theorem: Phase change in K-space = shift in image space
Convolution theorem: Multiply in K-space = filter in image space

</details>

💡 Plain English: What IS the Fourier Transform?

Imagine you're at a party with 100 people all talking at once. The Fourier Transform is like having a magical microphone that can:

Listen to the mixed noise (K-space data)

Figure out exactly who is saying what

Create a map showing where each person is standing (image!)

The math separates overlapping frequencies just like your ear separates different voices - but perfectly and mathematically!
K-space          →→→ FOURIER →→→        Image
(frequencies)        TRANSFORM          (spatial)

All mixed up                            Sorted by location
   ┌───┐                                  ┌───────┐
   │~~~│                                  │ █   █ │
   │~~~│            "Unmixing"            │   █   │
   │~~~│           frequencies            │ █ █ █ │
   └───┘                                  └───────┘

5.2 Practical Implementation

Discrete Fourier Transform (DFT)

Applied to digitised K-space data
Fast Fourier Transform (FFT) algorithms
Computational efficiency: O(N log N) vs O(N²)

Zero-filling (Interpolation)

Adding zeros to K-space periphery
Increases apparent matrix size
Improves image smoothness (not true resolution)

⚠️ Clinical Takeaway: Zero-filling Truth

Zero-filling does NOT create new information! It's like enlarging a photo - smoother pixels, but no new detail.

Original 256×256          Zero-filled to 512×512
┌─────────────┐           ┌─────────────────────┐
│ ██  ██  ██  │           │ ████    ████    ████│
│             │    →→→    │ ████    ████    ████│
│ ██  ██  ██  │           │                     │
└─────────────┘           │ ████    ████    ████│
                          │ ████    ████    ████│
Actual resolution:        └─────────────────────┘
Same!
                          Looks smoother, but same
                          actual detail level

When to use: Zero-filling is useful for smoother-looking images and reducing partial volume in reconstructions. Don't confuse it with true resolution improvement!

6. Clinical Applications and Protocol Optimisation

6.1 K-space Manipulation Techniques

Technique	K-space Strategy	Clinical Benefit
Partial Fourier	Sample >50% of K-space	Reduced scan time
Parallel Imaging	Under-sample with coil sensitivity	2-4× acceleration
Compressed Sensing	Random under-sampling	Ultra-fast imaging
Keyhole Imaging	Update central K-space only	Dynamic studies
PROPELLER/BLADE	Rotating strips through centre	Motion correction

⚠️ Clinical Takeaway: When to Use Each Technique

Clinical Scenario Best K-space Technique Why
Uncooperative patient PROPELLER/BLADE Motion correction from oversampled centre
Dynamic contrast study Keyhole Fast updates of contrast information
Breath-hold limited Parallel imaging Reduce scan time by 2-4×
Cardiac cine Compressed sensing Very fast, sparse data works well
High-resolution anatomy Full K-space Maximum detail, no acceleration artefacts

6.2 Parameter Optimisation

For High SNR:

Maximise central K-space sampling
Reduce bandwidth (longer readout)
Increase NEX/NSA
Use appropriate coil selection

For High Resolution:

Extend peripheral K-space coverage
Increase matrix size
Reduce FOV (maintain sampling density)
Consider zero-filling for smoothness

6.3 Artefact Recognition

K-space-related Artefacts:

Gibbs ringing: Insufficient high-frequency sampling
Aliasing: Violated Nyquist criterion
Motion ghosting: K-space inconsistency between lines
Spike artefact: Single corrupted K-space point

🔬 Worked Example 1: Bandwidth and SNR Calculation

Clinical Scenario: You're scanning a small FOV wrist study. The radiologist asks you to improve SNR. Your current bandwidth is 250 Hz/pixel.

Question: If you reduce bandwidth to 125 Hz/pixel, how much does SNR improve?

<details> <summary>Click to see step-by-step solution</summary>

Step 1: Recall the relationship

SNR ∝ 1/√(Bandwidth)

Step 2: Set up the ratio

SNR_new     √(BW_old)     √250
-------- = ----------- = ------
SNR_old     √(BW_new)     √125

Step 3: Calculate

√250 / √125 = √(250/125) = √2 ≈ 1.41

Show answer

Answer: SNR improves by a factor of 1.41 (41% increase)

Trade-offs to consider:

❌ Chemical shift artefact doubles (proportional to 1/BW)
❌ Minimum TE increases
✅ Better SNR for small structures

Clinical decision: Good choice for wrist (no fat-water interfaces critical), but might be problematic for spine (chemical shift at vertebral endplates).

</details>

🔬 Worked Example 2: Scan Time Calculation

Clinical Scenario: You need to reduce scan time for an anxious patient. Current protocol:

TR = 500 ms
Phase encoding steps = 256
NEX = 2

Question: What are your options to reduce scan time by half?

<details> <summary>Click to see step-by-step solution</summary>

Step 1: Calculate current scan time

Scan time = TR × N_phase × NEX
          = 500 ms × 256 × 2
          = 256,000 ms
          = 256 seconds (4 min 16 sec)

Step 2: Options to achieve 128 seconds (half time)

Option A: Reduce NEX to 1

New time = 500 × 256 × 1 = 128 sec ✓
Trade-off: SNR drops by √2 ≈ 30% reduction

Option B: Reduce phase steps to 128

New time = 500 × 128 × 2 = 128 sec ✓
Trade-off: Resolution in phase direction halved

Option C: Reduce TR to 250 ms

New time = 250 × 256 × 2 = 128 sec ✓
Trade-off: Changed contrast weighting, may not be achievable

Option D: Use Parallel Imaging (R=2)

New time = 500 × 256/2 × 2 = 128 sec ✓
Trade-off: Some SNR loss, potential aliasing artefacts

Best choice depends on clinical question:

Need contrast → Option A or D
Need coverage → Option A or D
Need resolution → Option A or D
SNR critical → Option B or C (if contrast acceptable)

</details>

🔬 Worked Example 3: Nyquist and Aliasing

Clinical Scenario: Your abdominal scan shows wrap-around artefact. The FOV is 32 cm, but the patient's abdomen is 40 cm wide.

Question: How do you fix this without increasing scan time?

<details> <summary>Click to see step-by-step solution</summary>

Step 1: Understand the problem

Current FOV: 32 cm
Patient width: 40 cm
Overhang: 8 cm (4 cm each side)

The 4 cm outside FOV "wraps" to opposite side

Step 2: Solutions

Option A: Increase FOV to 40+ cm

• Works, but may reduce resolution if matrix unchanged
• If matrix increased → longer scan time

Option B: Oversampling (No Phase Wrap / Fold-over suppression)

• Double FOV in phase direction (to 64 cm effective)
• Acquire double the lines
• Discard outer half after reconstruction
• NO scan time penalty (done automatically on most scanners)
• Slight SNR improvement from extra data

Option C: Swap phase and frequency directions

If patient is wider L-R than A-P:
• Make frequency direction L-R (no aliasing possible)
• Make phase direction A-P (shorter dimension)
• Caveat: Motion artefacts now propagate differently

Option D: Saturation bands

• Place saturation bands outside FOV
• Suppresses signal that would alias
• Increases SAR and slightly increases scan time

Best practice: Option B (oversampling) is usually best - it's automatic on modern scanners with no real penalty!

</details>

7. Advanced Concepts for Board Preparation

7.1 Echo Formation Strategies

Gradient Echo (GRE)

FID rephased by gradient reversal
T2* weighting predominates
Faster acquisition possible
Susceptibility sensitivity increased

Spin Echo (SE)

180° RF pulse refocuses spins
True T2 weighting achieved
Compensates for B₀ inhomogeneities
Longer minimum TE

💡 Plain English: Gradient Echo vs Spin Echo

Remember our marching band analogy?

Gradient Echo (GRE):

Reverses the gradient to bring dephased spins back together

Only corrects for gradient-induced dephasing

Field inhomogeneities (T2') still cause signal loss

Result: T2 weighting* (includes field effects)

Spin Echo (SE):

Uses a 180° RF pulse to flip all spins

Like telling marchers to turn around and walk back

Corrects for BOTH gradient AND field inhomogeneity dephasing

Result: True T2 weighting (intrinsic tissue property only)
Signal
  │
  │    GRE: Decays with T2* (faster)
  │    ╲
  │     ╲    SE: Decays with T2 (slower)
  │      ╲
  │       ╲────────────────────
  │        ╲                    ────────
  │         ╲___                        ────
  │              ───___
  │                    ───___
  └──────────────────────────────────────────▶ TE
        ↑
   Same tissue, different signal decay!

7.2 Multi-echo Acquisitions

Each echo samples different K-space region
Enables:
- T2 mapping (multiple TE values)
- Fat-water separation (Dixon techniques)
- Increased efficiency (EPI, FSE/TSE)

Examination Preparation Focus Points

Calculation Practice:
- Bandwidth = 1/(Dwell time × N_frequency)
- FOV = 2π/(γ × G × Sampling time)
- Scan time = TR × N_phase × NEX
Conceptual Understanding:
- Why does motion affect phase encoding more than frequency encoding?
- How does partial Fourier maintain image quality?
- What determines the minimum achievable TE in different sequences?
Clinical Correlation:
- Selecting appropriate bandwidth for different anatomical regions
- Optimising K-space sampling for specific pathology
- Troubleshooting image quality issues through K-space analysis

⚠️ Clinical Takeaway: Quick Reference Card

K-space Rules of Thumb:

If you need... Prioritise... Accept trade-off...
Better contrast Centre K-space Some edge blur
Better resolution Edge K-space Longer scan time
Shorter scan Partial K-space Some SNR/resolution loss
Motion correction Oversampled centre Longer reconstruction
Fast dynamics Keyhole/View sharing Temporal blurring

SNR Relationships:
SNR ∝ √(NEX) × √(Voxel volume) × 1/√(Bandwidth) × 1/√(Matrix)

Double NEX        → SNR × 1.41 (but 2× time)
Half bandwidth    → SNR × 1.41 (but more chemical shift)
Double voxel size → SNR × 2.83 (but less resolution)

Key References

Haacke EM, Brown RW, Thompson MR, Venkatesan R. Magnetic Resonance Imaging: Physical Principles and Sequence Design. 2nd ed. Wiley-Liss; 2014. Chapters 8-11.
Bernstein MA, King KF, Zhou XJ. Handbook of MRI Pulse Sequences. Academic Press; 2004. Part II: Signal Generation and Detection.
McRobbie DW, Moore EA, Graves MJ, Prince MR. MRI: From Picture to Proton. 3rd ed. Cambridge University Press; 2017. Chapters 6-7.
Paschal CB, Morris HD. K-space in the clinic. J Magn Reson Imaging. 2004;19(2):145-159.
Mezrich R. A perspective on K-space. Radiology. 1995;195(2):297-315.

Self-Assessment Questions

Physics Application: Calculate the receiver bandwidth required for a 256×256 matrix with a dwell time of 10 μs. How would halving this bandwidth affect SNR and chemical shift artefact?
K-space Analysis: Draw a K-space diagram showing the data distribution for:
- A T2-weighted FSE sequence with ETL=16
- A gradient echo sequence with partial Fourier (75% sampling)
- Label the contrast-determining and resolution-determining regions
Clinical Problem-Solving: A lumbar spine protocol shows excessive Gibbs artefact at CSF-cord interfaces. Propose three K-space-based solutions, explaining the trade-offs for each.
Advanced Concept: Explain why radial K-space sampling is more motion-robust than Cartesian sampling. Include discussion of how motion affects different K-space regions.

Summary

Free Induction Decay represents the fundamental signal in MRI, encoding spatial information through frequency and phase modulation under gradient control. K-space serves as the mathematical framework for data collection, with central regions determining contrast and peripheral regions defining resolution. The Fourier transform bridges the frequency and image domains, enabling visualisation of anatomical structures. Advanced practitioners must understand these relationships to optimise protocols, troubleshoot artefacts, and implement emerging techniques. This knowledge forms the foundation for all advanced MRI applications, from parallel imaging to compressed sensing, making it essential for board certification and expert practice.

💡 Final Plain English Summary

The Big Picture in 60 Seconds:

FID = The "ringing bell" signal after we excite protons

K-space = A "recipe book" where we store frequency information

Centre = basic ingredients (contrast, brightness)

Edges = decoration details (sharpness, fine lines)

Fourier Transform = The "magic decoder" that converts recipes into pictures

Gradients = The "joystick" steering us through K-space

Nyquist = Sample fast enough or get aliasing artefacts!

Everything in MRI comes back to these concepts. Master them, and pulse sequences, artefacts, and protocol optimisation all make sense!

← Lesson 1.1: History and Evolution of MRI Technology

That's the end of the free preview

58 more lessons come with a plan.

See plans Back to course overview

Wider Bandwidth	Narrower Bandwidth
✅ Less chemical shift artefact	❌ More chemical shift artefact
✅ Shorter TE possible	❌ Longer minimum TE
❌ Lower SNR	✅ Higher SNR
✅ Less distortion	❌ More susceptibility

Clinical Scenario	Best K-space Technique	Why
Uncooperative patient	PROPELLER/BLADE	Motion correction from oversampled centre
Dynamic contrast study	Keyhole	Fast updates of contrast information
Breath-hold limited	Parallel imaging	Reduce scan time by 2-4×
Cardiac cine	Compressed sensing	Very fast, sparse data works well
High-resolution anatomy	Full K-space	Maximum detail, no acceleration artefacts

If you need...	Prioritise...	Accept trade-off...
Better contrast	Centre K-space	Some edge blur
Better resolution	Edge K-space	Longer scan time
Shorter scan	Partial K-space	Some SNR/resolution loss
Motion correction	Oversampled centre	Longer reconstruction
Fast dynamics	Keyhole/View sharing	Temporal blurring

Lesson 1.5: Free Induction Decay (FID) and K-space Fundamentals

Learning Objectives

Introduction

1. Free Induction Decay: Physics and Signal Characteristics

1.1 Definition and Physical Basis

1.2 Signal Generation Mechanism

1.3 Mathematical Representation

📊 Visual: FID Decay Curve

2. Signal Detection and Digitisation

2.1 Receiver Coil Configuration

2.2 Analog-to-Digital Conversion

3. K-space: The Spatial Frequency Domain

3.1 Conceptual Framework

3.2 K-space Architecture

📊 Visual: K-space Centre vs Periphery

3.3 K-space Coordinates

4. Gradient-Controlled K-space Navigation

4.1 Frequency Encoding (Readout Gradient)

4.2 Phase Encoding

4.3 K-space Trajectories

📊 Visual: K-space Filling Patterns

5. Fourier Transform: From K-space to Image Space

5.1 Mathematical Principles

5.2 Practical Implementation

6. Clinical Applications and Protocol Optimisation

6.1 K-space Manipulation Techniques

6.2 Parameter Optimisation

6.3 Artefact Recognition

🔬 Worked Example 1: Bandwidth and SNR Calculation

🔬 Worked Example 2: Scan Time Calculation

🔬 Worked Example 3: Nyquist and Aliasing

7. Advanced Concepts for Board Preparation

7.1 Echo Formation Strategies

7.2 Multi-echo Acquisitions

Examination Preparation Focus Points

Key References

Self-Assessment Questions

Summary