Zobraziť v EDS

Generation of synthetic CT from MRI for MRI‐based attenuation correction of brain PET images using radiomics and machine learning

Uložené v:
Podrobná bibliografia
Názov: Generation of synthetic CT from MRI for MRI‐based attenuation correction of brain PET images using radiomics and machine learning
Autori: Hoseinipourasl, Amin, Hossein-Zadeh, Gholam-Ali, Sheikhzadeh, Peyman, Arabalibeik, Hossein, Alavijeh, Shaghayegh Karimi, Zaidi, Habib, Ay, Mohammad Reza
Zdroj: Medical Physics. 52:3772-3784
Informácie o vydavateľovi: Wiley, 2025.
Rok vydania: 2025
Predmety: Adult, Male, Attenuation correction, Radiomics, 616.0757, Brain / diagnostic imaging, attenuation correction, Magnetic Resonance Imaging, Machine Learning, machine learning, PET/MRI, Image Processing, Computer-Assisted / methods, radiomics, Positron-Emission Tomography, synthetic CT, Machine learning, Humans, Female, Tomography, X-Ray Computed, Synthetic CT
Popis: BackgroundAccurate quantitative PET imaging in neurological studies requires proper attenuation correction. MRI‐guided attenuation correction in PET/MRI remains challenging owing to the lack of direct relationship between MRI intensities and linear attenuation coefficients.PurposeThis study aims at generating accurate patient‐specific synthetic CT volumes, attenuation maps, and attenuation correction factor (ACF) sinograms with continuous values utilizing a combination of machine learning algorithms, image processing techniques, and voxel‐based radiomics feature extraction approaches.MethodsBrain MR images of ten healthy volunteers were acquired using IR‐pointwise encoding time reduction with radial acquisition (IR‐PETRA) and VIBE‐Dixon techniques. synthetic CT (SCT) images, attenuation maps, and attenuation correction factors (ACFs) were generated using the LightGBM, a fast and accurate machine learning algorithm, from the radiomics‐based and image processing‐based feature maps of MR images. Additionally, ultra‐low‐dose CT images of the same volunteers were acquired and served as the standard of reference for evaluation. The SCT images, attenuation maps, and ACF sinograms were assessed using qualitative and quantitative evaluation metrics and compared against their corresponding reference images, attenuation maps, and ACF sinograms.ResultsThe voxel‐wise and volume‐wise comparison between synthetic and reference CT images yielded an average mean absolute error of 60.75 ± 8.8 HUs, an average structural similarity index of 0.88 ± 0.02, and an average peak signal‐to‐noise ratio of 32.83 ± 2.74 dB. Additionally, we compared MRI‐based attenuation maps and ACF sinograms with their CT‐based counterparts, revealing average normalized mean absolute errors of 1.48% and 1.33%, respectively.ConclusionQuantitative assessments indicated higher correlations and similarities between LightGBM‐synthesized CT and Reference CT images. Moreover, the cross‐validation results showed the possibility of producing accurate SCT images, MRI‐based attenuation maps, and ACF sinograms. This might spur the implementation of MRI‐based attenuation correction on PET/MRI and dedicated brain PET scanners with lower computational time using CPU‐based processors.
Druh dokumentu: Article
Popis súboru: application/pdf
Jazyk: English
ISSN: 2473-4209
0094-2405
DOI: 10.1002/mp.17867
Prístupová URL adresa: https://pubmed.ncbi.nlm.nih.gov/40355942
https://research.rug.nl/en/publications/eb2fd18d-1f3a-4765-a5f2-c20c9f5bfab6
https://hdl.handle.net/11370/eb2fd18d-1f3a-4765-a5f2-c20c9f5bfab6
https://doi.org/10.1002/mp.17867
https://archive-ouverte.unige.ch/unige:185190
https://doi.org/10.1002/mp.17867
Rights: Wiley Online Library User Agreement
Prístupové číslo: edsair.doi.dedup.....936c567a879797720e73416ded01a6a3
Databáza: OpenAIRE
Popis
Abstrakt:BackgroundAccurate quantitative PET imaging in neurological studies requires proper attenuation correction. MRI‐guided attenuation correction in PET/MRI remains challenging owing to the lack of direct relationship between MRI intensities and linear attenuation coefficients.PurposeThis study aims at generating accurate patient‐specific synthetic CT volumes, attenuation maps, and attenuation correction factor (ACF) sinograms with continuous values utilizing a combination of machine learning algorithms, image processing techniques, and voxel‐based radiomics feature extraction approaches.MethodsBrain MR images of ten healthy volunteers were acquired using IR‐pointwise encoding time reduction with radial acquisition (IR‐PETRA) and VIBE‐Dixon techniques. synthetic CT (SCT) images, attenuation maps, and attenuation correction factors (ACFs) were generated using the LightGBM, a fast and accurate machine learning algorithm, from the radiomics‐based and image processing‐based feature maps of MR images. Additionally, ultra‐low‐dose CT images of the same volunteers were acquired and served as the standard of reference for evaluation. The SCT images, attenuation maps, and ACF sinograms were assessed using qualitative and quantitative evaluation metrics and compared against their corresponding reference images, attenuation maps, and ACF sinograms.ResultsThe voxel‐wise and volume‐wise comparison between synthetic and reference CT images yielded an average mean absolute error of 60.75 ± 8.8 HUs, an average structural similarity index of 0.88 ± 0.02, and an average peak signal‐to‐noise ratio of 32.83 ± 2.74 dB. Additionally, we compared MRI‐based attenuation maps and ACF sinograms with their CT‐based counterparts, revealing average normalized mean absolute errors of 1.48% and 1.33%, respectively.ConclusionQuantitative assessments indicated higher correlations and similarities between LightGBM‐synthesized CT and Reference CT images. Moreover, the cross‐validation results showed the possibility of producing accurate SCT images, MRI‐based attenuation maps, and ACF sinograms. This might spur the implementation of MRI‐based attenuation correction on PET/MRI and dedicated brain PET scanners with lower computational time using CPU‐based processors.
ISSN:24734209
00942405
DOI:10.1002/mp.17867