A super-resolution feasibility study in small-animal SPECT imaging

Villena, Jose L.; Lage, Eduardo; Carlos, Alvaro de; Tapias, G.; Sisniega, Alejandro; Vaquero, J.J.; Desco, Manuel

doi:10.1109/nssmic.2008.4774306

Cited by 5 publications

(5 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The normal distribution was used to randomly generate phantoms in batches. The number of point sources in the six triangular regions in the Derenzo-like phantoms was random and could be selected from the integer range (1)(2)(3)(4)(5)(6)(7)(8)(9). The intensity values of the background and point sources were random decimals and could be selected from the range (0.0-1.0).…”

Section: Digital Phantoms and Preprocessingmentioning

confidence: 99%

“…One of the most important strategies is to apply multi-image SR technology on the image domain (3). In the projection domain, an HR projection can be calculated from a series of LR projections containing different information, and then the image is reconstructed using the reconstruction method (4)(5)(6). However, the design of these methods is based on a perfect, error-free calibration system, and, in practice, errors in motion and registration may introduce uncertainty.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Super-resolution reconstruction for parallel-beam SPECT based on deep learning and transfer learning: a preliminary simulation study

Cheng¹,

Wen²,

Zhang³

et al. 2022

Ann Transl Med

View full text Add to dashboard Cite

Background: Single-photon emission computed tomography (SPECT) is widely used in the early diagnosis of major diseases such as cardiovascular disease and cancer. High-resolution (HR) imaging requires HR projection data, which typically comes with high costs. This study aimed to obtain HR SPECT images based on a deep learning algorithm using low-resolution (LR) detectors.Methods: A super-resolution (SR) reconstruction network based on deep learning and transfer learning for parallel-beam SPECT was proposed. LR SPECT sinograms were converted into HR sinograms.Training data were designed and generated, including digital phantoms (128×128 pixels), HR sinograms (128×128 pixels), and LR sinograms (128×64 pixels). A series of random phantoms was first used for pretraining, and then the extended cardiac-torso (XCAT) phantom was used to fine-tune the parameters. The effectiveness of the method was evaluated using an unknown cardiac phantom. To simulate a wide range of noise levels, the total count levels of the projection were normalized to 1e7 (100%), 1e6 (10%), and 1e5 (5%). Finally, the training sets for different count levels were generated. Transfer learning was employed to accelerate the training. Results:The proposed network was validated on the simulation data sets using different Poisson noise levels. The quantitative values of the peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) indicators of the reconstructed images were improved compared to those recorded using the benchmark methods. Using the proposed method, an image resolution comparable to that of images reconstructed from the HR projection could be achieved.Conclusions: Based on deep learning and transfer learning, an SR reconstruction network in the projection domain of the parallel-beam SPECT was developed. The simulation results under a wide range of noise levels evidenced the potential of the proposed network to improve SPECT resolution for LR detector scanners.

show abstract

Section: Digital Phantoms and Preprocessingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Super-resolution reconstruction for parallel-beam SPECT based on deep learning and transfer learning: a preliminary simulation study

Cheng¹,

Wen²,

Zhang³

et al. 2022

Ann Transl Med

View full text Add to dashboard Cite

show abstract

“…This method ow computational cost even in SR techniques have been widely used in many applications, such as surveillance, vision-based forensic, remote sensing, and satellite imaging. In medical imaging, there have been some works applying on Magnetic Resonance Imaging (MRI) [23], Computed tomography (CT) [24], Single-Photon Emission Computed Tomography (SPECT) [25] and Positron Emission Tomography (PET) [26], [27].…”

Section: Introductionmentioning

confidence: 99%

Dose reduction in breast imaging: A phantom study

Kantharak

Chalidabhongse

Prueksadee

2015

2015 2nd International Conference on Biomedical Engineering (ICoBE)

View full text Add to dashboard Cite

This paper proposed a method in breast imaging. While mammogram is a early detection of breast cancer, it still limi women. So, an additional investigation usin Tomosynthesis (DBT) is then required. This m of radiation that patients received is approxim propose a method to reduce the radiation do synthesized 2D mammographic image from Super-Resolution (SR) technique. In this p estimation algorithm is performed within a (ROI) and then the estimated motion param reconstruct a SR image. The synthesized 2 image is the result of SR reconstruction step. method using the Peak signal to noise ratio structure similarity (MSSIM). The results 0.9976 respectively. In additions, we also ev using the phantom passing score. Our mammographic image got 5, 3, 4 for microcalcifications and masses, respectively.

show abstract

“…22 Others have incorporated a super-resolution approach directly on a set of SPECT sinograms obtained by shifting the gamma cameras. 23 The approach is similar to the wobble approaches in Refs. 4 and 5, which involve filtered backprojection ͑FBP͒ from sinograms with increased sampling density; however, in Ref.…”

Section: Introductionmentioning

confidence: 99%

“…4 and 5, which involve filtered backprojection ͑FBP͒ from sinograms with increased sampling density; however, in Ref. 23, FBP is applied to the high resolution sinogram obtained by applying a super-resolution algorithm on the different low-resolution sinograms.…”

Section: Introductionmentioning

confidence: 99%

A PET supersets data framework for exploitation of known motion in image reconstruction

Verhaeghe¹,

Reader²

2010

Medical Physics

View full text Add to dashboard Cite

The proposed superset framework allows comprehensive analysis of the commonly adopted approximations when considering motion in PET reconstruction. The findings demonstrate that the supersets data format successfully unify reconstruction in the presence of different sources of known acquisition motion into one framework, and as a result leads to hitherto unreported image quality improvements for all the cases tested where the object motion is accurately known.

show abstract

A super-resolution feasibility study in small-animal SPECT imaging

Cited by 5 publications

References 6 publications

Super-resolution reconstruction for parallel-beam SPECT based on deep learning and transfer learning: a preliminary simulation study

Super-resolution reconstruction for parallel-beam SPECT based on deep learning and transfer learning: a preliminary simulation study

Dose reduction in breast imaging: A phantom study

A PET supersets data framework for exploitation of known motion in image reconstruction

Contact Info

Product

Resources

About