"Compressed sensing is an advanced MRI reconstruction technique that accelerates image acquisition by undersampling k-space data and reconstructing images using sparsity and iterative algorithms 1. It enables faster scans, reduced motion artefacts, and improved patient comfort, while maintaining diagnostic image quality 3."
"incoherent undersampling: random undersampling of k-space spreads artefacts as noise rather than structured aliasing"
"potential artefacts: residual undersampling artefacts or “noise-like” patterns may be introduced if acceleration is excessive"
"potential artefacts: residual undersampling artefacts or “noise-like” patterns may be introduced if acceleration is excessive"
"k-space data are acquired with pseudo-random sampling patterns 1,3"
"Compressed sensing was developed in the mid-2000s by Emmanuel Candès, Terence Tao, and David DonohoEarly CS-MRI implementations appeared in 2007–2008 research publications 1. Commercial integration into MRI scanners began in the mid-2010s, with cardiac MRI leading the adoption."
"sparsity: MR images are sparse in certain transform domains (e.g. wavelets, discrete cosine)"
"TE, TR, and flip angle remain sequence dependent; compressed sensing primarily alters k-space coverage and reconstruction, not inherent tissue contrast"
"TE, TR, and flip angle remain sequence dependent; compressed sensing primarily alters k-space coverage and reconstruction, not inherent tissue contrast"
"often combined with parallel imaging (SENSE, GRAPPA) and/or deep learning reconstruction to further accelerate scans 1-3"
"often combined with parallel imaging (SENSE, GRAPPA) and/or deep learning reconstruction to further accelerate scans 1-3"
Expected headings
"Limitations"
"cardiac MRI: shortens breath-hold sequences; improves dynamic imaging (cine, perfusion, late gadolinium enhancement) 4"
"neuroimaging: accelerates 3D structural imaging (e.g. MPRAGE, FLAIR); used in angiography (time-of-flight MRA)"
"TE, TR, and flip angle remain sequence dependent; compressed sensing primarily alters k-space coverage and reconstruction, not inherent tissue contrast"
"neuroimaging: accelerates 3D structural imaging (e.g. MPRAGE, FLAIR); used in angiography (time-of-flight MRA)"
"sparsity: MR images are sparse in certain transform domains (e.g. wavelets, discrete cosine)"