ALL-Net: Anatomical information lesion-wise loss function integrated into neural network for multiple sclerosis lesion segmentation

Zhang, Hang; Zhang, Jinwei; Li, Chao; Sweeney, Elizabeth; Spincemaille, Pascal; Nguyen, Thanh D.; Gauthier, Susan A.; Wang, Yi; Marcille, Melanie

doi:10.1016/j.nicl.2021.102854

Cited by 26 publications

(22 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…43 To minimize effects of this undesired variation, future in vivo validation should include automated lesion segmentation techniques. [44][45][46][47] Nevertheless, our data suggest that our approximation produces a good qualitative agreement with independent measurements of iron and myelin within MS lesions. for specific quantitative mapping of myelin 49,50 and laser ablation inductively coupled plasma mass spectroscopy for iron quantitative mapping.…”

Section: Resultssupporting

confidence: 73%

“…In the present study, lesion segmentation for correlation analysis was performed manually on T2 FLAIR images, an approach prone to intra‐ and interreader variability 43 . To minimize effects of this undesired variation, future in vivo validation should include automated lesion segmentation techniques 44–47 . Nevertheless, our data suggest that our approximation produces a good qualitative agreement with independent measurements of iron and myelin within MS lesions.…”

Section: Discussionmentioning

confidence: 85%

See 1 more Smart Citation

Susceptibility source separation from gradient echo data using magnitude decay modeling

Dimov

Nguyen

Gillen

et al. 2022

Journal of Neuroimaging

Self Cite

View full text Add to dashboard Cite

Background and Purpose:The objective is to demonstrate feasibility of separating magnetic sources in quantitative susceptibility mapping (QSM) by incorporating magnitude decay rates R * 2 in gradient echo (GRE) MRI. Methods: Magnetic susceptibility source separation was developed using R * 2 and compared with a prior method using R ′ 2 = R * 2 − R 2 that required an additional sequence to measure the transverse relaxation rate R 2 . Both susceptibility separation methods were compared in multiple sclerosis (MS) patients (n = 17). Susceptibility values of negative sources estimated with R * 2 -based source separation in a set of enhancing MS lesions (n = 44) were correlated against longitudinal myelin water fraction (MWF) changes. Results:In in vivo data, linear regression of the estimated 𝜒 + and 𝜒 − susceptibility values between the R * 2 -and the R ′ 2 -based separation methods performed across 182 segmented lesions revealed correlation coefficient r = .96 and slope close .99. Correlation analysis in enhancing lesions revealed a significant positive association between the 𝜒 − increase at 1-year post-onset relative to 0 year and the MWF increase at 1 year relative to 0 year (β = -0.144, 95% confidence interval: [-0.199, -0.1], p = .0008) and good agreement between R ′ 2 and R * 2 methods (r = .79, slope = .95). Conclusions: Separation of magnetic sources based solely on GRE complex data is feasible by combining magnitude decay rate modeling and phase-based QSM and 𝜒 − change may serve as a biomarker for myelin recovery or damage in acute MS lesions.

show abstract

Section: Resultssupporting

confidence: 73%

Section: Discussionmentioning

confidence: 85%

Susceptibility source separation from gradient echo data using magnitude decay modeling

Dimov

Nguyen

Gillen

et al. 2022

Journal of Neuroimaging

Self Cite

View full text Add to dashboard Cite

show abstract

“…2). Future work involves pairing QSMRim-Net with an automated T2-FLAIR lesion segmentation algorithm, such as All-Net [21] with geometric loss [54] and attention-based approaches [55] [56], followed by an automated method to separate confluents lesions [57]. We plan to adapt and train the algorithm to work directly on T2-FLAIR lesion segmentations.…”

Section: Discussionmentioning

confidence: 99%

“…In studies of chronic active MS lesions on MRI, lesions are typically identified on the T2-weighted-Fluid-Attenuated Inversion Recovery (T2-FLAIR) image and then are determined to be chronic active through visual inspection on susceptibility imaging. This process is time consuming and prone to inter- and intra-rater variability [20, 21]. For these lesions to be further studied at a large scale and translated into clinical practice, there is a great need for automated methods to identify chronic active MS lesions.…”

Section: Introductionmentioning

confidence: 99%

QSMRim-Net: Imbalance-Aware Learning for Identification of Chronic Active Multiple Sclerosis Lesions on Quantitative Susceptibility Maps

Zhang

Nguyen

Zhang

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Background and Purpose: Chronic active multiple sclerosis (MS) lesions are characterized by a paramagnetic rim at the edge of the lesion and are associated with increased disability in patients. Quantitative susceptibility mapping (QSM) is an MRI technique that is sensitive to chronic active lesions, termed rim+ lesions on the QSM. We present QSMRim-Net, a data imbalance-aware deep neural network that fuses lesion-level radiomic and convolutional image features for automated identification of rim+ lesions on QSM. Methods: QSM and T2-weighted-Fluid-Attenuated Inversion Recovery (T2-FLAIR) MRI of the brain were collected at 3T for 172 MS patients. Rim+ lesions were manually annotated by two human experts, followed by consensus from a third expert, for a total of 177 rim+ and 3986 rim negative (rim-) lesions. Our automated rim+ detection algorithm, QSMRim-Net, consists of a two-branch feature extraction network and a synthetic minority oversampling network to classify rim+ lesions. The first network branch is for image feature extraction from the QSM and T2-FLAIR, and the second network branch is a fully connected network for QSM lesion-level radiomic feature extraction. The oversampling network is designed to increase classification performance with imbalanced data. Results: On a lesion-level, in a five-fold cross validation framework, the proposed QSMRim-Net detected rim+ lesions with a partial area under the receiver operating characteristic curve (pROC AUC) of 0.760, where clinically relevant false positive rates of less than 0.1 were considered. The method attained an area under the precision recall curve (PR AUC) of 0.704. QSMRim-Net out-performed other state-of-the-art methods applied to the QSM on both pROC AUC and PR AUC. On a subject-level, comparing the predicted rim+ lesion count and the human expert annotated count, QSMRim-Net achieved the lowest mean square error of 0.98 and the highest correlation of 0.89 (95% CI: 0.86, 0.92). Conclusion: This study develops a novel automated deep neural network for rim+ MS lesion identification using T2-FLAIR and QSM images.

show abstract

“…Convolutional neural networks (CNNs) have been the dominant approach across a variety of medical image computing applications such as magnetic resonance imaging (MRI) reconstruction [Zhang et al, 2020c;Zhang et al, 2021b], brain lesion segmentation [Zhang et al, 2021a;Zhang et al, 2019a], disease identification [Khosla et al, 2019;Zhang et al, 2022], and low-dose CT denoising Fan et al, 2019]. Two main properties [Elsayed et al, 2020] of CNNs are believed to be the key to their success: 1) The local receptive field increases as the network goes deeper [Huang et al, 2017;He et al, 2016], providing richer contextual information for down-stream tasks; 2) Spatially invariant convolution filters regularize the network training, serving as a good inductive bias for grid-based image data [Simoncelli and Olshausen, 2001].…”

Section: Introductionmentioning

confidence: 99%

Spatially Covariant Lesion Segmentation

Zhang¹,

Wang²,

Zhang³

et al. 2023

Preprint

View full text Add to dashboard Cite

Compared to natural images, medical images usually show stronger visual patterns and therefore this adds flexibility and elasticity to resource-limited clinical applications by injecting proper priors into neural networks. In this paper, we propose spatially covariant pixel-aligned classifier (SCP) to improve the computational efficiency and meantime maintain or increase accuracy for lesion segmentation. SCP relaxes the spatial invariance constraint imposed by convolutional operations and optimizes an underlying implicit function that maps image coordinates to network weights, the parameters of which are obtained along with the backbone network training and later used for generating network weights to capture spatially covariant contextual information. We demonstrate the effectiveness and efficiency of the proposed SCP using two lesion segmentation tasks from different imaging modalities: white matter hyperintensity segmentation in magnetic resonance imaging and liver tumor segmentation in contrast-enhanced abdominal computerized tomography. The network using SCP has achieved 23.8%, 64.9% and 74.7% reduction in GPU memory usage, FLOPs, and network size with similar or better accuracy for lesion segmentation.

show abstract

ALL-Net: Anatomical information lesion-wise loss function integrated into neural network for multiple sclerosis lesion segmentation

Cited by 26 publications

References 39 publications

Susceptibility source separation from gradient echo data using magnitude decay modeling

Susceptibility source separation from gradient echo data using magnitude decay modeling

QSMRim-Net: Imbalance-Aware Learning for Identification of Chronic Active Multiple Sclerosis Lesions on Quantitative Susceptibility Maps

Spatially Covariant Lesion Segmentation

Contact Info

Product

Resources

About