Improving Sensitivity of Arterial Spin Labeling Perfusion <scp>MRI</scp> in Alzheimer's Disease Using Transfer Learning of Deep Learning‐Based <scp>ASL</scp> Denoising

Zhang, Lei; Xie, Dong; Li, Yiran; Camargo, Aldo; Song, Donghui; Lu, Tian; Jeudy, Jean; Dreizin, David; Melhem, Elias R.; Wang, Ze; Initiative, Alzheimer’s Disease Neuroimaging

doi:10.1002/jmri.27984

Cited by 19 publications

(18 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several classifiers were added to the end of FC layers and obtained an accuracy of 88–100% for different classifiers such as KNN, Naïve Bayes, SVM, and Logistic Regression on top of AlexNet and GoogLeNet. Liang et al ( 2018 ), Eitel et al ( 2019 ), Khan et al ( 2019 ), Oh et al ( 2019 ), Puranik et al ( 2019 ), Ramzan et al ( 2020 ), Simon et al ( 2019 ), Wu et al ( 2019 ), Zhang et al ( 2021 ), and Ocasio and Duong ( 2021 ) also utilized this approach and obtained competitive results.…”

Section: Resultsmentioning

confidence: 99%

Transfer Learning Approaches for Neuroimaging Analysis: A Scoping Review

Ardalan

Subbian

2022

Front. Artif. Intell.

View full text Add to dashboard Cite

Deep learning algorithms have been moderately successful in diagnoses of diseases by analyzing medical images especially through neuroimaging that is rich in annotated data. Transfer learning methods have demonstrated strong performance in tackling annotated data. It utilizes and transfers knowledge learned from a source domain to target domain even when the dataset is small. There are multiple approaches to transfer learning that result in a range of performance estimates in diagnosis, detection, and classification of clinical problems. Therefore, in this paper, we reviewed transfer learning approaches, their design attributes, and their applications to neuroimaging problems. We reviewed two main literature databases and included the most relevant studies using predefined inclusion criteria. Among 50 reviewed studies, more than half of them are on transfer learning for Alzheimer's disease. Brain mapping and brain tumor detection were second and third most discussed research problems, respectively. The most common source dataset for transfer learning was ImageNet, which is not a neuroimaging dataset. This suggests that the majority of studies preferred pre-trained models instead of training their own model on a neuroimaging dataset. Although, about one third of studies designed their own architecture, most studies used existing Convolutional Neural Network architectures. Magnetic Resonance Imaging was the most common imaging modality. In almost all studies, transfer learning contributed to better performance in diagnosis, classification, segmentation of different neuroimaging diseases and problems, than methods without transfer learning. Among different transfer learning approaches, fine-tuning all convolutional and fully-connected layers approach and freezing convolutional layers and fine-tuning fully-connected layers approach demonstrated superior performance in terms of accuracy. These recent transfer learning approaches not only show great performance but also require less computational resources and time.

show abstract

Section: Resultsmentioning

confidence: 99%

Transfer Learning Approaches for Neuroimaging Analysis: A Scoping Review

Ardalan

Subbian

2022

Front. Artif. Intell.

View full text Add to dashboard Cite

show abstract

“…Denote y ∈ C N (C is the complex domain, and N is the dimension) as the desired image to be restored, which consists of √ N × √ N pixels, where the undersampled k-space data is represented by x ∈ C M , in which M << N. Denote the linear operator (undersampling process) with F eter selections. Outside of the MR research field, the past several years have seen sensational success from a variety of model-free deep neural networks in image classification [8], computer vision, auditory processing, information generation, and medical imaging [9][10][11][12]. Motivated by such superb performance of deep learning (DL) [13], different groups have incorporated various DL networks into MRI reconstructions such as a deep convolutional autoencoder network [14], deep residual learning [15], deep ADMM-net [16], U-Net [17], and Cycle-GAN [18].…”

Section: Problem Definition and Notationsmentioning

confidence: 99%

“…For a nonlinear undersampling operator, the corresponding image reconstruction process is often formulated to be an optimization problem: ∈ C M×N , which can be further represented by F eter selections. Outside of the MR research field, the pa tional success from a variety of model-free deep neural [8], computer vision, auditory processing, information g [9][10][11][12]. Motivated by such superb performance of deep le have incorporated various DL networks into MRI recon lutional autoencoder network [14], deep residual learnin Net [17], and Cycle-GAN [18].…”

Section: Problem Definition and Notationsmentioning

confidence: 99%

“…2021, 11, x FOR PEER REVIEW eter selections. Outside of the MR research field, the past several years tional success from a variety of model-free deep neural networks in ima [8], computer vision, auditory processing, information generation, and [9][10][11][12]. Motivated by such superb performance of deep learning (DL) [13], have incorporated various DL networks into MRI reconstructions such lutional autoencoder network [14], deep residual learning [15], deep AD Net [17], and Cycle-GAN [18].…”

Section: Problem Definition and Notationsmentioning

confidence: 99%

See 1 more Smart Citation

POCS-Augmented CycleGAN for MR Image Reconstruction

Yang

Xie

et al. 2021

Applied Sciences

Self Cite

View full text Add to dashboard Cite

Recent years have seen increased research interest in replacing the computationally intensive Magnetic resonance (MR) image reconstruction process with deep neural networks. We claim in this paper that the traditional image reconstruction methods and deep learning (DL) are mutually complementary and can be combined to achieve better image reconstruction quality. To test this hypothesis, a hybrid DL image reconstruction method was proposed by combining a state-of-the-art deep learning network, namely a generative adversarial network with cycle loss (CycleGAN), with a traditional data reconstruction algorithm: Projection Onto Convex Set (POCS). The output of the first iteration’s training results of the CycleGAN was updated by POCS and used as the extra training data for the second training iteration of the CycleGAN. The method was validated using sub-sampled Magnetic resonance imaging data. Compared with other state-of-the-art, DL-based methods (e.g., U-Net, GAN, and RefineGAN) and a traditional method (compressed sensing), our method showed the best reconstruction results.

show abstract

“…Kim et al [9] first introduced a CNN-based network for ASL denoising with local and global pathways. Xie et al [7] developed a Dilated Wide Activation Network (DWAN) for denoising ASL images which has been applied to 2D pulsed ASL (PASL) data in persons with mild cognitive impairment and Alzheimer's disease [7], [10]. Gong et al [8] introduced multi-contrast input including ASL and M0 images to the network to further improve the performance for denoising 3D pCASL data.…”

Section: Introductionmentioning

confidence: 99%

Transformer based deep learning denoising of single and multi-delay 3D Arterial Spin Labeling

Shou

Zhao

Shao

et al. 2023

Preprint

View full text Add to dashboard Cite

Purpose To present a Swin Transformer-based deep learning (DL) model for denoising of single-delay and multi-delay 3D arterial spin labeling (ASL) data and compare its performance with convolutional neural network (CNN) methods. Methods Transformer and CNN-based spatial denoising models were developed for single-delay ASL. The models were trained on 2 datasets (59 subjects, 104 scans) and tested on 5 datasets (44 subjects, 57 scans) of single-delay ASL including data from 3 different vendors. Spatiotemporal denoising models were further developed using one dataset (6 subjects, 10 scans) of multi-delay ASL. A range of 2D, pseudo-3D and pseudo-4D input conditions for denoising single and multi-delay ASL data respectively was tested. The performance was evaluated using similarity metrics (SSIM and PSNR), spatial signal-to-noise ratio (SNR) and quantification accuracy of cerebral blood flow (CBF) and arterial transit time (ATT). Results Swin Transformer outperformed CNN-based networks, whereas pseudo-3D models showed better performance than standard 2D models for denoising single-delay ASL data. The similarity metrics (SSIM and PSNR) and image quality (SNR) improved with larger number of slices in pseudo-3D models which further improved when the M0 image was included as input, but at the cost of greater biases for CBF quantification. Pseudo-3D models with 3 slices input achieved optimal balance between SNR and accuracy, which can be generalized to ASL data from different vendors. For multi-delay ASL, spatiotemporal denoising models based on Swin Transformer had better performance compared to CNN or spatial-only models with reduced biases in fitted CBF and ATT maps. Conclusions By taking into account spatiotemporal information in ASL data, Swin Transformer DL models provided improved performance compared to CNN methods for denoising both single and multi-delay 3D ASL data. The proposed model offers flexibility to improve image quality and/or reduce scan time for 3D ASL to facilitate its clinical use.

show abstract

Improving Sensitivity of Arterial Spin Labeling Perfusion MRI in Alzheimer's Disease Using Transfer Learning of Deep Learning‐Based ASL Denoising

Cited by 19 publications

References 32 publications

Transfer Learning Approaches for Neuroimaging Analysis: A Scoping Review

Transfer Learning Approaches for Neuroimaging Analysis: A Scoping Review

POCS-Augmented CycleGAN for MR Image Reconstruction

Transformer based deep learning denoising of single and multi-delay 3D Arterial Spin Labeling

Contact Info

Product

Resources

About