Search

Background and Aims: 18F-fluoro-ethyl-tyrosine (FET) is a radiopharmaceutical used in positron emission tomography (PET)-computed tomography in patients with glioma. We propose an original approach combining a radiotracer-pharmacokinetic exploration performed at the voxel level (three-dimensional pixel) and voxel classification to identify tumor tissue. Our methodology was validated using the standard FET-PET approach and magnetic resonance imaging (MRI) data acquired according to the current clinical practices.Methods: FET-PET and MRI data were retrospectively analyzed in ten patients presenting with progressive high-grade glioma. For FET-PET exploration, radioactivity acquisition started 15 min after radiotracer injection, and was measured each 5 min during 40 min. The tissue segmentation relies on population pharmacokinetic modeling with dependent individuals (voxels). This model can be approximated by a linear mixed-effects model. The tumor volumes estimated by our approach were compared with those determined with the current clinical techniques, FET-PET standard approach (i.e., a cumulated value of FET signal is computed during a time interval) and MRI sequences (T1 and T2/fluid-attenuated inversion recovery [FLAIR]), used as references. The T1 sequence is useful to identify highly vascular tumor and necrotic tissues, while the T2/FLAIR sequence is useful to isolate infiltration and edema tissue located around the tumor.Results: With our kinetic approach, the volumes of tumor tissue were larger than the tissues identified by the standard FET-PET and MRI T1, while they were smaller than those determined with MRI T2/FLAIR.Conclusion: Our results revealed the presence of suspected tumor voxels not identified by the standard PET approach. [ABSTRACT FROM AUTHOR]

Purpose: Respiratory motion is a source of artifacts that reduce image quality in PET. Four dimensional (4D) PET/CT is one approach to overcome this problem. Existing techniques to limiting the effects of respiratory motions are based on prospective phase binning which requires a long acquisition duration (15-25 min). This time is uncomfortable for the patients and limits the clinical exploitation of 4D PET/CT. In this work, the authors evaluated an existing method and an alternative retrospective binning method to reduce the acquisition duration of 4D PET/CT. Methods: The authors studied an existing mixed-amplitude binning (MAB) method and an alternative binning method by mixed-phases (MPhB). Before implementing MPhB, they analyzed the regularity of the breathing patterns in patients. They studied the breathing signal drift and missing CT slices that could be challenging for implementing MAB. They compared the performance of MAB and MPhB with current binning methods to measure the maximum uptake, internal volume, and maximal range of tumor motion. Results: MPhB can be implemented depending on an optimal phase (in average, the exhalation peak phase -4.1% of the entire breathing cycle duration). Signal drift of patients was in average 35% relative to the breathing amplitude. Even after correcting this drift, MAB was feasible in 4D CT for only 64% of patients. No significant differences appeared between the different binning methods to measure the maximum uptake, internal volume, and maximal range of tumor motion. The authors also determined the inaccuracies of MAB and MPhB to measure the maximum amplitude of tumor motion with three bins (less than 3 mm for movement inferior to 12 mm, up to 6.4 mm for a 21 mm movement). Conclusions: The authors proposed an alternative binning method by mixed-phase binning that halves the acquisition duration of 4D PET/CT. Mixed-amplitude binning was challenging because of signal drift and missing CT slices. They showed that more than three bins were necessary for a more accurate measurement of the maximum amplitude of the tumor motion. However, the current 4D-CT technology limits the increase of the number of bins in 4D PET/CT because of missing CT slices. One can reconstruct 4D PET images with more bins but without attenuation/scatter correction. [ABSTRACT FROM AUTHOR]

Simple Summary: Due to the central role of magnetic resonance Imaging (MRI) in the management of patients with cancer, waiting lists exceed clinically relevant delays. For this reason, many research groups and MRI manufacturers develop algorithms as resampling and denoising models to allow faster acquisition time without deterioration in image quality. Whereas these algorithms are available in all new MRI, it is not clear how they will impact image features as well as the validity of statistical model of radiomics which use deep images characteristics to predict treatment outcome. The aim of this study was to develop resampling and denoising deep learning (DL) models and evaluate their impact on radiomics from post-Gd-T1w-MRI brain images with brain metastases. We show that resampling and denoising DL models reconstruct low resolution and noised MRI images acquired quickly into high quality images. While fast acquisition loses most of the radiomic-features and invalidates predictive radiomic models, DL models restore these parameters. Background: Magnetic resonance imaging (MRI) is predominant in the therapeutic management of cancer patients, unfortunately, patients have to wait a long time to get an appointment for examination. Therefore, new MRI devices include deep-learning (DL) solutions to save acquisition time. However, the impact of these algorithms on intensity and texture parameters has been poorly studied. The aim of this study was to evaluate the impact of resampling and denoising DL models on radiomics. Methods: Resampling and denoising DL model was developed on 14,243 T1 brain images from 1.5T-MRI. Radiomics were extracted from 40 brain metastases from 11 patients (2049 images). A total of 104 texture features of DL images were compared to original images with paired t-test, Pearson correlation and concordance-correlation-coefficient (CCC). Results: When two times shorter image acquisition shows strong disparities with the originals concerning the radiomics, with significant differences and loss of correlation of 79.81% and 48.08%, respectively. Interestingly, DL models restore textures with 46.15% of unstable parameters and 25.96% of low CCC and without difference for the first-order intensity parameters. Conclusions: Resampling and denoising DL models reconstruct low resolution and noised MRI images acquired quickly into high quality images. While fast MRI acquisition loses most of the radiomic features, DL models restore these parameters. [ABSTRACT FROM AUTHOR]