Reliability of SUV estimates in FDG PET as a function of acquisition and processing protocols

Feuardent, Juliette; Soret, Marine; Dreuille, O. De; Fœhrenbach, H.; Buvat, Irène

doi:10.1109/nssmic.2003.1352484

Cited by 6 publications

(6 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The large majority of these algorithms, which have been evaluated mostly in the context of cerebral imaging, require a segmentation step to delineate the different parts from anatomical images. This step renders their accuracy dependent on the segmentation algorithm used as well as making their application on other clinical investigations outside the brain challenging (Feuardent et al 2003). For example, the pixel-based approach of Meltzer et al (1990) is restricted indeed to brain metabolism or neuroreceptor binding, and requires compartmental analysis (Meltzer et al 1999).…”

Section: Introductionmentioning

confidence: 99%

A multiresolution image based approach for correction of partial volume effects in emission tomography

et al. 2006

View full text Add to dashboard Cite

Partial volume effects (PVEs) are consequences of the limited spatial resolution in emission tomography. They lead to a loss of signal in tissues of size similar to the point spread function and induce activity spillover between regions. Although PVE can be corrected for by using algorithms that provide the correct radioactivity concentration in a series of regions of interest (ROIs), so far little attention has been given to the possibility of creating improved images as a result of PVE correction. Potential advantages of PVE-corrected images include the ability to accurately delineate functional volumes as well as improving tumour-to-background ratio, resulting in an associated improvement in the analysis of response to therapy studies and diagnostic examinations, respectively. The objective of our study was therefore to develop a methodology for PVE correction not only to enable the accurate recuperation of activity concentrations, but also to generate PVE-corrected images. In the multiresolution analysis that we define here, details of a high-resolution image H (MRI or CT) are extracted, transformed and integrated in a low-resolution image L (PET or SPECT). A discrete wavelet transform of both H and L images is performed by using the "à trous" algorithm, which allows the spatial frequencies (details, edges, textures) to be obtained easily at a level of resolution common to H and L. A model is then inferred to build the lacking details of L from the high-frequency details in H. The process was successfully tested on synthetic and simulated data, proving the ability to obtain accurately corrected images. Quantitative PVE correction was found to be comparable with a method considered as a reference but limited to ROI analyses. Visual improvement and quantitative correction were also obtained in two examples of clinical images, the first using a combined PET/CT scanner with a lymphoma patient and the second using a FDG brain PET and corresponding T1-weighted MRI in an epileptic patient.

show abstract

Section: Introductionmentioning

confidence: 99%

A multiresolution image based approach for correction of partial volume effects in emission tomography

et al. 2006

View full text Add to dashboard Cite

show abstract

“…The resulting value varies according to biological factors (glycemia, steroids, uptake period duration, injected activity) and reconstruction parameters. 11 In conclusion, indolent lymphoma patients showing HT symptoms should systematically undergo FDG-PET/CT investigation before biopsies are performed. Biopsies should be performed in the SUVmax site.…”

mentioning

confidence: 99%

Investigation of FDG-PET/CT imaging to guide biopsies in the detection of histological transformation of indolent lymphoma

Bodet-Milin¹,

Kraeber-Bodéré²,

Moreau³

et al. 2008

Haematologica

130

View full text Add to dashboard Cite

Positron emission tomography using 18F-fluorodeoxyglucose (FDG-PET) has been successfully evaluated in the management of non-Hodgkin's lymphoma (NHL).1-3 Histological transformation (HT) of indolent lymphoma is a dramatic event that occurs in 5-10% of the patients and carries a dismal prognosis.4,5 Previous studies prove that indolent lymphoma entities show a lower FDG uptake when compared with aggressive lymphomas.6-8 We therefore postulated that FDG-PET/CT identifies aggressive transformation sites and can guide biopsies.

show abstract

“…In clinic, the mean and maximum SUVs are often used as bases for diagnosis to characterize suspicious high uptakes [31,32]. Therefore, the SUV measures are used to investigate the…”

Section: Clinical Evaluation For Specific Voismentioning

confidence: 99%

“…In general, physicians use both the maximum SUV (SUV max ) and the mean SUV (SUV mean ) to characterize the high uptakes [31], but SUV max is more often used in practice since SUV mean heavily depends on volume of interest (VOI) selected, while SUV max value is unique and reproducible in VOI [32,33]. Inspired by the most recent advanced neural networks, such as Dense-Net [34], Residual CNN [35], and CycleGAN [36], a cycle Wasserstein regression adversarial training framework, named S-CycleGAN, is proposed and studied for the PET brain imaging in this paper.…”

Section: Introductionmentioning

confidence: 99%

Study of low-dose PET image recovery using supervised learning with CycleGAN

Zhao¹,

Zhou²,

Gao³

et al. 2020

PLoS ONE

View full text Add to dashboard Cite

PET is a popular medical imaging modality for various clinical applications, including diagnosis and image-guided radiation therapy. The low-dose PET (LDPET) at a minimized radiation dosage is highly desirable in clinic since PET imaging involves ionizing radiation, and raises concerns about the risk of radiation exposure. However, the reduced dose of radioactive tracers could impact the image quality and clinical diagnosis. In this paper, a supervised deep learning approach with a generative adversarial network (GAN) and the cycle-consistency loss, Wasserstein distance loss, and an additional supervised learning loss, named as S-CycleGAN, is proposed to establish a non-linear end-to-end mapping model, and used to recover LDPET brain images. The proposed model, and two recently-published deep learning methods (RED-CNN and 3D-cGAN) were applied to 10% and 30% dose of 10 testing datasets, and a series of simulation datasets embedded lesions with different activities, sizes, and shapes. Besides vision comparisons, six measures including the NRMSE, SSIM, PSNR, LPIPS, SUV max and SUV mean were evaluated for 10 testing datasets and 45 simulated datasets. Our S-CycleGAN approach had comparable SSIM and PSNR, slightly higher noise but a better perception score and preserving image details, much better SUV mean and SUV max , as compared to RED-CNN and 3D-cGAN. Quantitative and qualitative evaluations indicate the proposed approach is accurate, efficient and robust as compared to other stateof-the-art deep learning methods.

show abstract

Reliability of SUV estimates in FDG PET as a function of acquisition and processing protocols

Cited by 6 publications

References 4 publications

A multiresolution image based approach for correction of partial volume effects in emission tomography

A multiresolution image based approach for correction of partial volume effects in emission tomography

Investigation of FDG-PET/CT imaging to guide biopsies in the detection of histological transformation of indolent lymphoma

Study of low-dose PET image recovery using supervised learning with CycleGAN

Contact Info

Product

Resources

About