• Rasika jaywant sonawane
• Torana kamble

: Magnetic Resonance Imaging (MRI) provides excellent soft-tissue contrast but lacks the quantitative Hounsfield Unit (HU) information that Computed Tomography (CT) provides. Accurate CT data are essential for radiation therapy dose planning, attenuation correction in PET/MR, and surgical navigation. To reduce the need for dual-modality scanning, researchers have developed algorithms to predict CT images directly from MRI. One important direction is feature matching with learned nonlinear local descriptors, where local MRI patterns are encoded into discriminative descriptors, matched to reference datasets, and mapped to CT values through nonlinear transformations. This paper presents a comprehensive literature survey of MRI-to-CT prediction methods with particular emphasis on approaches that leverage learned nonlinear local descriptors for feature matching and pseudo-CT synthesis. The aim is to highlight the strengths, limitations, and clinical potential of these emerging techniques.

MRI, CT, Deep learning

"A Survey on Predicting CT Image from MRI Data through Feature Matching", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 10, page no.a336-a339, October-2025, Available :http://www.jetir.org/papers/JETIR2510042.pdf

"A Survey on Predicting CT Image from MRI Data through Feature Matching", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 10, page no. ppa336-a339, October-2025, Available at : http://www.jetir.org/papers/JETIR2510042.pdf