2.3 PET/US: Attenuation Correction & Denoising

This chapter connects two practical preprocessing topics:

PET attenuation correction (AC) for quantitative imaging
Ultrasound noise/artifacts and basic denoising strategies

PET: why attenuation correction is essential

Gamma photons are attenuated (absorbed/scattered) inside the body, causing systematic underestimation for deeper structures.

Common approaches:

PET-CT: build a μ-map from CT (and map energies to 511 keV)
PET-MRI: estimate μ-map via segmentation/atlas/UTE (harder)

AC is one part of a full quantitative chain (scatter/randoms/dead-time/normalization are also common).

Ultrasound: noise and artifacts

Typical issues:

Speckle (coherent imaging phenomenon)
Shadowing / enhancement
Reverberation
Refraction / side lobes

Denoising strategies:

classical edge-preserving filters (median, bilateral, diffusion)
learning-based/self-supervised denoisers (task-aware)