Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

Motion Correction of Multi-Frame PET Data

  • Cañizares, Gabriel
  • Espinós-Morató, Héctor
  • Hernández, Ángel
  • Álvarez-Gómez, Juan Manuel
  • Santos-Blasco, Joaquín
  • Moliner Martínez, Laura
  • González Martínez, Antonio Javier
  • F Sánchez
  • Rodríguez-Álvarez, M.J.|||0000-0001-8333-8792
  • Benlloch Baviera, Jose María|||0000-0001-6073-1436
[EN] Motion correction algorithms are necessary in PET studies where the patient cannot remain totally static. Otherwise, artifacts appear in the final reconstructed image, degrading the spatial resolution or even misplacing the radiation distribution source. In this work we propose a new enhanced motion correction algorithm (EMAF) based on the multiple acquisition frames (MAF). This methodology can be applied to two PET system geometries (a ring and a multi-panel system). Using simulated data, the comparison of static and motion corrected profiles shows minimum differences keeping more than 99% of the events after the process, with a minimal inter-frame error (the maximum range of loss counts is around 0.6%). To measure the quality of the correction method, two error metrics as peak signal to noise ratio (PSNR) and intensity matching precision (IMP) are proposed. A Mini-Derenzo in a ring PET system reveals a remarkable improvement both for these two metrics (44.5 dB and 98.7%) when compared to the uncorrected images (41.1 dB and 56%, respectively). An equivalent analysis for a point-like source also shows a sizeable enhancement both in contrast and resolution (64 dB and 98% versus 62.5 dB and 17.9%, PSNR and IMP correspondingly). The proposed algorithm minimizes the image artifacts and its simplicity, independency of PET configuration system and rapid reconstruction and registration times, makes it a useful tool in preclinical PET studies., This work has been co-financed by the Spanish Government grants TEC2016-79884-C2 and RTC-2016-5186-1, and by the European Union through the European Regional Development Fund (ERDF)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

Deep learning for MRI-based CT synthesis: a comparison of MRI sequences and neural network architectures

  • Larroza, Andrés
  • Álvarez-Gómez, Juan Manuel
  • Espinós-Morató, Héctor
  • Vergara-Díaz, Marina
  • Moliner Martínez, Laura
  • Oliver-Gil, Sandra|||0000-0001-8258-3972
  • Rodríguez-Álvarez, M.J.|||0000-0001-8333-8792
[EN] Synthetic computed tomography (CT) images derived from magnetic resonance images (MRI) are of interest for radiotherapy planning and positron emission tomography (PET) attenuation correction. In recent years, deep learning implementations have demonstrated improvement over atlasbased and segmentation-based methods. Nevertheless, several open questions remain to be addressed, such as which is the best of MRI sequences and neural network architectures. In this work, we compared the performance of different combinations of two common MRI sequences (T1- and T2-weighted), and three state-of-the-art neural networks designed for medical image processing (Vnet, HighRes3dNet and ScaleNet). The experiments were conducted on brain datasets from a public database. Our results suggest that T1 images perform better than T2, but the results further improve when combining both sequences. The lowest mean average error over the entire head (MAE = 101.76 ± 10.4 HU) was achieved combining T1 and T2 scans with HighRes3dNet. All tested deep learning models achieved significantly lower MAE (p < 0.01) than a well-known atlas-based method., This work was supported by the Spanish Government grants TEC2016-79884-C2 and RTC-2016-5186-1, and by the European Union through the European Regional Development Fund (ERDF)