177
Part Four · Quantum biology

Quantum imaging and MRI physics

A hospital MRI scanner is a quantum spin experiment the size of a room. Protons in your body’s water behave as spin-½ magnets (Chapter 40); a 3-tesla field splits their two states by an energy so small that at body temperature only about one proton in 100,000 is net aligned. From that whisper of polarization, radio pulses, magnetic field gradients and a Fourier transform build images of soft tissue with millimetre detail. As a biomedical engineer you may already use MRI; this chapter shows the physics underneath, and where quantum techniques could push it further.

On the plate: magnetization tipped, precessing and relaxing.

Precession and polarization

In a field B0B_0, proton magnetization precesses at the Larmor frequency f=γB0/2πf = \gamma B_0/2\pi, with γ/2π=42.58\gamma/2\pi = 42.58 MHz/T: 127.7 MHz at 3 T. The equilibrium polarization is P≈γℏB0/2kBTP\approx\gamma\hbar B_0/2k_BT, about 10−510^{-5} at 3 T and 310 K. Water’s huge proton density, about 6.7×10286.7\times10^{28} per cubic metre, rescues the signal. Felix Bloch and Edward Purcell independently detected nuclear magnetic resonance in condensed matter in 1946 and shared the 1952 Nobel Prize in Physics. Established

Bloch equations, T₁ and T₂

A 90° radio pulse tips the magnetization into the transverse plane, where it precesses and induces a voltage in a coil. Two relaxations follow, described by Bloch’s equations: the transverse signal decays with time constant T2T_2 as spins dephase, and the longitudinal magnetization regrows toward equilibrium with time constant T1T_1 (the same concepts as for qubits, Chapter 135). Set T1=T_1 = 900 ms and T2=T_2 = 100 ms; grey matter at 3 T is roughly 1,300 and 100 ms, fat about 380 and 130 ms. Image contrast comes from choosing pulse timings that weight T1T_1 or T2T_2. Field inhomogeneity makes the signal decay faster still; Erwin Hahn’s 1950 spin echo, a 180° pulse that reverses the dephasing, recovers it. Established

k-space

Paul Lauterbur showed in 1973 that a magnetic field gradient makes precession frequency depend on position, so the signal at each moment is one Fourier component of the image. The scanner fills k-space, the 2D Fourier transform of the image, line by line, and an inverse transform recovers the picture. Choose which part of k-space to keep: . The centre carries contrast; the edges carry fine detail; skipping lines makes the image fold over (aliasing). Fast MRI methods, from parallel imaging to compressed sensing, exploit these trade-offs.

Hyperpolarization attacks the 10⁻⁵ problem directly. Jan Henrik Ardenkjær-Larsen and colleagues showed in 2003 that dissolution dynamic nuclear polarization, transferring electron polarization to nuclei at about 1 K and then dissolving, raises carbon-13 polarization over ten-thousandfold, enabling real-time imaging of metabolism, such as pyruvate conversion in tumours. Quantum-enhanced imaging with entangled or squeezed light, ghost imaging and sub-shot-noise microscopy, works in optics labs but has not yet changed clinical imaging. Established for MRI and DNP.

What you now have

Larmor precession and tiny thermal polarization, Bloch equations with T₁ and T₂, spin echoes, gradients and k-space, how sampling choices create blur and aliasing, hyperpolarization, and the status of quantum-enhanced imaging.

· · ·

Three questions to carry with you.

Why is the MRI signal so weak per proton?tap
The Zeeman splitting is tiny compared with k_BT, so polarization is about 10⁻⁵.
What does the centre of k-space encode?tap
Large-scale contrast; the edges encode fine detail.
How does a spin echo work?tap
A 180° pulse reverses dephasing from static field variations, refocusing the signal.

And one to work out. With γ/2π=42.58\gamma/2\pi = 42.58 MHz/T, what is the proton Larmor frequency at 3 T, in MHz?

f =

Sources: F. Bloch, Phys. Rev. 70, 460 (1946). E. M. Purcell, H. C. Torrey and R. V. Pound, Phys. Rev. 69, 37 (1946). E. L. Hahn, Phys. Rev. 80, 580 (1950). P. C. Lauterbur, Nature 242, 190 (1973). J. H. Ardenkjær-Larsen et al., Proc. Natl Acad. Sci. USA 100, 10158 (2003). Checked by code: P = γħB/2k_BT = 9.9 × 10⁻⁶ at 3 T, 310 K; the plate reconstructs a 32 × 32 phantom by exact inverse DFT.

Next · Chapter 178
Quantum computing for drug discovery and protein folding (overview; detail in Part 5)

AlphaFold won a Nobel Prize without any quantum computer. So where would quantum help?

Turn the page