Edge-Enhanced Dilated Residual Attention Network for Multimodal Medical Image Fusion

Farzad Khalvati; Jiayi Wang; Meng Zhou; Xiaolan Xu; Yuxuan Zhang

REVIEW 5 major objections 4 minor 14 references

Edge-Enhanced Dilated Residual Attention Network for Multimodal Medical Image Fusion

T0 review · 5 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read An edge-enhanced dilated residual attention network with a parameter-free softmax-nuclear-norm fusion rule reports the best MRI-CT and MRI-SPECT fusion metrics among six methods and lifts brain tumor classification AUC from 0.727 to 0.769.

desk verdict A workmanlike incremental fusion paper whose reported results are likely fine, but the fusion rule as written is undefined and the overclaims need correction before the numbers can be trusted. read the letter →

arxiv 2411.11799 v1 pith:DSPAHLJY submitted 2024-11-18 eess.IV cs.AIcs.CV

Meng Zhou , Yuxuan Zhang , Xiaolan Xu , Jiayi Wang , Farzad Khalvati This is my paper · ORCID

classification eess.IVcs.AIcs.CV

keywords medicalimagefusionmultimodaldilatedresidualattentionnetworkedgeenhancementsoftmaxfeatureweightedstrategyparameter-freenuclearnormgliomaclassification

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper aims to show that a purely convolutional fusion framework can rival heavier Transformer-based models on multimodal medical image fusion while remaining fast enough for clinical use. It proposes an asymmetric autoencoder whose encoder uses dilated residual attention blocks for multi-scale features and a gradient operator for edge details, with a parameter-free fusion rule that weights feature maps through the softmax-transformed nuclear norm. On MRI-CT and MRI-SPECT test pairs, the method reports the highest PSNR, FMI, FSIM, and entropy among the six compared methods, with fusion in about one second per pair. A downstream low-grade versus high-grade glioma (LGG/HGG) classification experiment on fused T2-FLAIR images improves AUC over FLAIR alone (0.769 versus 0.727), which the authors offer as evidence that fusion quality carries clinical signal.

What carries the argument

The load-bearing object is the fusion rule: Softmax Feature Weighted Strategy (SFNN), a parameter-free scheme in which each latent feature map is transformed channel-wise by softmax, reduced to a matrix nuclear norm, aggregated across channels with $\phi(\cdot)$ (set to max in the best reported variant), and normalized across modalities to produce weights for a weighted sum of the feature maps. This rule is what lets the paper claim real-time fusion because it contributes no trainable parameters and no extra computation in training or inference. The two named architectural components that carry the feature-quality gains are the Dilated Residual Attention Network (DRAN), which stacks convolutional groups at increasing depths to emulate $3\times3$, $5\times5$, and $7\times7$ receptive fields and adds dilated convolutions with rates $\{1,3,5\}$ to widen the effective field without downsampling, and the Dense Residual Gradient Operator (DRGO), which adds a learnable Sobel-gradient magnitude term to convolutional features in the shallow layers. Together they supply the multi-scale and edge information that the fusion rule then weighs.

What would settle it

Re-run the MRI-CT and MRI-SPECT comparisons with a separate validation split used to pick $\phi(\cdot)$ among mean, sum, and max, compute the published metrics only once on the untouched test set, and check whether SFNN-max still exceeds FER and FL1N; if it does not, the central claim of a superior fusion strategy fails.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that a deliberately asymmetric CNN autoencoder—a deep encoder built from dilated residual attention and a dense residual gradient operator, paired with a shallow decoder—extracts latent feature maps rich enough that a parameter-free fusion rule can produce fused images that outperform recent CNN and Transformer baselines on MRI-CT and MRI-SPECT fusion, and that such fused images improve brain tumor type classification. The fusion rule, called the Softmax Feature Weighted Strategy (SFNN) with max aggregation, takes each latent feature map through a channel-wise softmax, computes the matrix nuclear norm of the softmax output, normalizes those norms across modalities, and uses the resulting scalar weights to form a weighted sum of feature maps; it adds no trainable parameters and no extra inference cost. The headline numbers are the highest PSNR, FMI, FSIM, and entropy on MRI-CT (16.830, 0.883, 0.820, 10.727) and the highest PSNR, SSIM, FMI, and FSIM on MRI-SPECT (21.455, 0.736, 0.876, 0.843), and a fused T2-FLAIR input that raises low- vs high-grade glioma (LGG/HGG) classification AUC to 0.769 from 0.727 for FLAIR alone.

Load-bearing premise

The load-bearing premise is that the test sets were untouched until the final evaluation, when in fact the fusion-strategy variant max was chosen by comparing mean, sum, and max on those same test sets; if the test sets had been used only once, the reported advantage over alternative parameter-free strategies could shrink or disappear.

Editorial extensions

If this is right

If the reported fusion metrics hold, a CNN-based pipeline can fuse MRI-CT or MRI-SPECT pairs in roughly one second, which is fast enough for interactive clinical use.
Because the fusion rule is parameter-free, the fusion stage adds no learnable parameters or inference-time cost; any speed gain or loss comes entirely from the encoder and decoder.
The ablation results attribute the quality gain to both the gradient loss and the edge-enhancement module, so edge-preserving training is a direct cause of the reported SSIM and FSIM improvements.
The downstream classification result implies that fused T2-FLAIR images carry more LGG/HGG discriminative information than either single modality or a two-channel concatenation, making fusion a viable preprocessing step for automated tumor classification.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the two-modality experiments, the SFNN formula is written for $C$ images, so the same parameter-free rule could fuse three or more modalities without retraining the fusion module; whether that scales is untested in the paper.
A reader could test the edge-enhancement hypothesis by comparing fusion quality on anatomy-rich versus edge-sparse regions: the method's advantage over baselines should be largest where edges are densest.
The classification gain is reported on a single dataset with one classifier and three random splits; confirming the clinical value would require a multi-center cohort and a comparison against expert reading time.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The paper is a reasonably serious attempt at CNN-based medical image fusion: the network combines residual attention, pyramid attention, dilated convolutions, and a Sobel gradient branch, which is a sensible incremental architecture, and they validate it on a downstream LGG/HGG classification task, which is a nice extra. The quantitative improvements over the five baselines (especially PSNR and FSIM on MRI-CT and MRI-SPECT) are plausible and roughly in line with the qualitative examples. If the implementation matches the text, this is useful progress in a crowded subfield.

That said, there is a load-bearing problem. Equation (2) defines the fusion weights as W_k = phi(||S(x_i)_k||_*) / sum_k phi(||S(x_i)_k||_*), and the text says S(x_i)_k is the softmax weight map for modality k. That object is a matrix, so its nuclear norm is a scalar, and phi = identity, mean, sum, or max on a scalar all give the same value. That makes SFNN-mean, SFNN-sum, and SFNN-max identical, which is contradicted by Table 2, where they report different numbers. The only way this makes sense is if phi is intended to aggregate a vector of per-channel nuclear norms, but that operation is not in the equation or the surrounding text. As written, the central fusion strategy is not a well-defined algorithm, and the headline results in Table 1 are tied to an undefined variant. This needs to be fixed before anything else.

There are also some smaller soft spots. The text claims "fusion in only 1 second" while Table 3 gives 1.26s, slower than three CNN baselines; the ablation text claims improvements across all metrics, but MRI-SPECT SSIM actually drops (0.743 to 0.736) when the gradient loss and DRGO are added; and the choice of SFNN-max appears to have been made after evaluating on the same test sets used for the final numbers, with no validation split. None of these are fatal on their own, but they compound the reproducibility concern, and the paper does not release code yet.

The paper deserves a serious referee because the architecture and the downstream experiment are worth engaging with, but the review should be heavy. The authors need to rewrite the fusion equation, provide a validation split or clearly state how SFNN-max was chosen, correct the speed and ablation statements, and release code. Until then, the numbers should be treated with skepticism.

Referee Report

5 major / 4 minor

Summary. The paper proposes EH-DRAN, a two-stage asymmetric autoencoder for multimodal medical image fusion (MRI-CT and MRI-SPECT). Stage 1 trains an encoder-decoder with a Dilated Residual Attention Network and a Dense Residual Gradient Operator under pixel, gradient, and perceptual losses; Stage 2 fuses the latent feature maps with a parameter-free Softmax Feature Weighted Strategy based on the nuclear norm of softmax weight maps, with φ chosen as max. The authors report improved fusion metrics over five baselines, an ablation of the edge enhancer and gradient loss, and a downstream BraTS LGG/HGG classification experiment where a fused T2-FLAIR image yields AUC 0.769 versus 0.727 for FLAIR alone.

Significance. If the technical issues are resolved, the contribution is positive: the training protocol is a self-contained reconstruction task, the fusion strategy is parameter-free, and the downstream classification evaluation is a useful step beyond pixel-level fusion metrics. The paper also deserves credit for reporting standard deviations over multiple test sets and for making the code release explicit. However, the central fusion-strategy definition in Eq. (2) is internally inconsistent with the reported experiments, so the significance of the current version is contingent on a corrected formulation.

major comments (5)

[Section 2.2, Eq. (2)] Equation (2) defines W_k = φ(||S(x_i)_k||_*) / Σ_k φ(||S(x_i)_k||_*). Since ||S(x_i)_k||_* is a single nonnegative scalar for each modality k, applying φ ∈ {identity, mean, sum, max} to that scalar returns the same value, so W_a/W_b is identical for all four variants and the fused feature map is identical. Table 2 nonetheless reports different PSNR, SSIM, FMI, FSIM, and EN values for SFNN-mean, SFNN-sum, and SFNN-max on both datasets. The authors must specify the vector over which φ aggregates (for example, per-channel nuclear norms or the singular values of the weight map), correct Eq. (2) accordingly, and rerun the fusion-strategy comparison and all subsequent results that depend on SFNN-max.
[Section 4.1, 'Fusion Time' and Table 3] The text states that the proposed method 'achieves image fusion in only 1 second', but Table 3 reports an average inference time of 1.26 seconds on MRI-SPECT, which is slower than IFCNN (0.89 s), MSRPAN (0.79 s), and MSDRA (0.81 s). This contradicts the speed claim in the abstract and in the introduction. Please either correct the timing statement, report the actual measured time, or remove the sub-second claim.
[Section 4.1, Ablation Study and Table 4] The ablation text says that adding L_grad and then the DRGO module led to 'improvements across all evaluation metrics' compared with the base model. Table 4 shows this is not the case: for MRI-SPECT, SSIM decreases monotonically from 0.743 (Base Model) to 0.740 (Base Model + L_grad) to 0.736 (full model), and for MRI-CT the FMI gain from adding DRGO is marginal (0.881 to 0.883). The narrative should be revised to describe the actual metric-by-metric changes rather than claiming universal improvement.
[Section 4.1, 'Selection of Fusion Strategies' and Table 2] The fusion strategy φ is selected as max after evaluating mean, sum, and max on the same standalone test sets (30 MRI-CT pairs and 50 MRI-SPECT pairs) that are later used for the final comparisons in Table 1. Since no validation split is used to choose φ, the reported best numbers may be inflated by selection on the evaluation data. Please introduce a validation split for strategy selection, or explicitly report the selection procedure and its limitations.
[Section 4.2, Table 5 and classification discussion] The text claims that the fused image 'significantly improves performance in terms of AUC and F1-Score', but no statistical significance test is reported. In Table 5, the F1-Score for the fused image (0.723 ± 0.006) overlaps with the T2+FLAIR result (0.717 ± 0.012), and Accuracy is identical (0.640 ± 0.011 versus 0.640 ± 0.015). Only AUC shows a clear separation (0.769 ± 0.003 versus 0.727 ± 0.024). Please add significance tests or temper the claim accordingly.

minor comments (4)

[Abstract] The sentence 'requires no additional computations during training or inference' is inaccurate because the fusion strategy computes softmax operations and nuclear norms at inference; please rephrase as 'no additional learned parameters' or 'no training-time optimization'.
[Section 2.2] The citation (Lahoud & Süsstrunk, 2019) appears in the text for the idea of weight maps, but this reference is missing from the reference list; please add it.
[Introduction] The phrase 'the emergency of the improved selective structured state space models' should be 'the emergence of ...'; this appears to be a typographical error.
[Equation (3) and surrounding text] The text introduces M as the number of input images, but M does not appear in the displayed loss formula; please clarify the notation or remove the unused symbol.

Circularity Check

2 steps flagged · score 6.0 of 10

The fusion-strategy claim partially reduces to its own test-set selection, and Eq. (2) makes the SFNN phi-variants identical by construction, so the reported SFNN-max advantage is not independently derived.

self definitional [Section 2.2, Eq. (2); Section 4.1, Table 2]
"Wk = ϕ(∥S(xi)k∥∗)PC k=1 ϕ(∥S(xi)k∥∗) ... As mentioned, we selected ϕ(·) to be max() in this work, but it can also be mean(), sum() or identity()."

In Eq. (2), phi is applied to the scalar nuclear norm ||S(x_i)_k||_*. For any non-negative scalar t, identity(t), mean(t), sum(t), and max(t) all equal t. With C=2, the normalized weights W_a and W_b are therefore identical under every listed phi, and the final fused map f = sum_k W_k f^k is the same map for SFNN-identity, SFNN-mean, SFNN-sum, and SFNN-max. Table 2 nevertheless reports different PSNR, SSIM, FMI, FSIM, and EN for SFNN-mean, SFNN-sum, and SFNN-max on both datasets. Hence the distinct 'max' fusion strategy is not defined by the stated equations; the claimed variant superiority is an artifact of an unspecified operation, not a consequence of the formula in the paper.
fitted input called prediction [Section 3 test protocol; Section 4.1, Table 2 and Table 1]
"We randomly held out 30 image pairs from the MRI-CT dataset and 50 pairs from the MRI-SPECT dataset as the standalone test set. ... We conducted an extensive qualitative and quantitative analysis of these strategies for both the MRI-CT and MRI-SPECT fusion tasks, with the results summarized in Table 2."

The same standalone test sets used to compare the SFNN variants in Table 2 are then reused to report the final EH-DRAN numbers in Table 1, and the EH-DRAN row in Table 1 is numerically identical to the SFNN-max row in Table 2. Thus the choice of phi = max is evaluated on the test data, and the headline claim that 'our proposed method outperforms' is a test-set-fitted selection rather than an independent prediction. No validation split is described between this fusion-strategy selection and the final metric reporting, so the comparison to baselines is statistically forced by the selection procedure.

full rationale

Most of the pipeline is self-contained: the Stage-1 autoencoder is trained with reconstruction losses on real images without fusion labels, and the downstream brain-tumor classification uses a fixed ResNet-50 on external BraTS data, with the Zhou & Khalvati (2024) citation used only for the training setup rather than as load-bearing evidence. However, the central fusion-strategy claim has two construction-level problems. First, Eq. (2) defines phi on a scalar nuclear norm, so identity, mean, sum, and max are mathematically identical; the different SFNN rows in Table 2 cannot be produced by the stated formula, making the 'max' variant collapse by construction into a single unspecified rule. Second, the phi = max choice is selected on the same test sets later used for the headline Table 1 numbers, with the EH-DRAN row equal to the SFNN-max row, so the reported superiority is a test-set-fitted result rather than an independent prediction. These are partial circularities in the fusion-strategy claim; the encoder/decoder reconstruction itself is not circular.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim depends on a few empirical choices: loss weights, a data-filtering threshold, and the assumption that reconstruction-trained features transfer to fusion. There are no invented physical or mathematical entities; all components are adapted from prior work.

free parameters (3)

lambda1 (gradient loss weight) = 0.5
Empirically set in Section 2.3; affects texture detail learning in the reconstructed images.
lambda2 (perceptual loss weight) = 0.5
Empirically set in Section 2.3; affects semantic similarity weighting.
minimum non-zero pixel fraction for slice selection = 10%
Used to filter axial slices from BraTS; changes the training distribution for the downstream classification task.

assumptions (3)

domain assumption Fusion quality can be measured by PSNR, SSIM, FMI, FSIM and entropy computed between the fused image and source images, despite the absence of a ground truth fused image.
Metrics in Section 3 follow prior fusion literature, but PSNR and SSIM semantics for fusion without a reference are debated.
domain assumption An autoencoder trained only for reconstruction learns latent features that are informative for multimodal fusion.
Stage-1 reconstruction training is separated from fusion; the paper assumes these latent features transfer to the fusion task, see Section 2.1.
domain assumption The softmax-weighted nuclear-norm rule is a valid parameter-free fusion operator.
The rule is proposed in Section 2.2 without a derivation from optimality principles; its validity rests on empirical performance.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Edge-Enhanced Dilated Residual Attention Network for Multimodal Medical Image Fusion." pith.science (2026). https://pith.science/paper/DSPAHLJY

@misc{pith2026241111799,
  author       = {Pith},
  title        = {Pith review of: Edge-Enhanced Dilated Residual Attention Network for Multimodal Medical Image Fusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DSPAHLJY}},
  note         = {Machine review of arXiv:2411.11799}
}

read the original abstract

Multimodal medical image fusion is a crucial task that combines complementary information from different imaging modalities into a unified representation, thereby enhancing diagnostic accuracy and treatment planning. While deep learning methods, particularly Convolutional Neural Networks (CNNs) and Transformers, have significantly advanced fusion performance, some of the existing CNN-based methods fall short in capturing fine-grained multiscale and edge features, leading to suboptimal feature integration. Transformer-based models, on the other hand, are computationally intensive in both the training and fusion stages, making them impractical for real-time clinical use. Moreover, the clinical application of fused images remains unexplored. In this paper, we propose a novel CNN-based architecture that addresses these limitations by introducing a Dilated Residual Attention Network Module for effective multiscale feature extraction, coupled with a gradient operator to enhance edge detail learning. To ensure fast and efficient fusion, we present a parameter-free fusion strategy based on the weighted nuclear norm of softmax, which requires no additional computations during training or inference. Extensive experiments, including a downstream brain tumor classification task, demonstrate that our approach outperforms various baseline methods in terms of visual quality, texture preservation, and fusion speed, making it a possible practical solution for real-world clinical applications. The code will be released at https://github.com/simonZhou86/en_dran.

Figures

Figures reproduced from arXiv: 2411.11799 by the authors.

**Figure 2.** A sample illustration of the fusion process, we take MRI-SPECT fusion as an example. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Qualitative results for MRI-CT (top three rows) and MRI-SPECT (bottom three rows) [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: For the MRI-CT fusion task, our method demonstrated clear advantages over the FER [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

**Figure 4.** Figure 4: Qualitative comparison between different fusion strategies on MRI-CT and SPECT [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 3 canonical work pages

[4]

Furthermore, compared to the FL1N method (Li et al., 2022), our approach produced brighter and more defined edges, thereby improving the contrast between edges and tissues

For the MRI-CT fusion task, our method demonstrated clear advantages over the FER strategy (Fu et al., 2021), achieving enhanced fidelity with sharper delineation of inner tissue struc- tures from MRI images and well-preserved boundaries from CT images. Furthermore, compared to the FL1N method (Li et al., 2022), our approach produced brighter and more def...

work page 2021
[8]

Weisheng Li, Xiuxiu Peng, Jun Fu, Guofen Wang, Yuping Huang, and Feifei Chao

doi: 10.1109/TIP.2018.2887342. Weisheng Li, Xiuxiu Peng, Jun Fu, Guofen Wang, Yuping Huang, and Feifei Chao. A multiscale double-branch residual attention network for anatomical–functional medical image fusion. Com- puters in Biology and Medicine, 141:105005,

arXiv 2018
[10]

The multimodal brain tumor image segmentation benchmark (brats)

Bjoern H Menze, Andras Jakab, Stefan Bauer, Jayashree Kalpathy-Cramer, Keyvan Farahani, Justin Kirby, Yuliya Burren, Nicole Porz, Johannes Slotboom, Roland Wiest, et al. The multimodal brain tumor image segmentation benchmark (brats). IEEE transactions on medical imaging , 34 (10):1993–2024,

1993
[11]

Fusionmamba: Efficient image fusion with state space model

Siran Peng, Xiangyu Zhu, Haoyu Deng, Zhen Lei, and Liang-Jian Deng. Fusionmamba: Efficient image fusion with state space model. arXiv preprint arXiv:2404.07932,

arXiv
[13]

Xinyu Xie, Yawen Cui, Chio-In Ieong, Tao Tan, Xiaozhi Zhang, Xubin Zheng, and Zitong Yu

doi: 10.1109/TIM.2023.3317470. Xinyu Xie, Yawen Cui, Chio-In Ieong, Tao Tan, Xiaozhi Zhang, Xubin Zheng, and Zitong Yu. Fusionmamba: Dynamic feature enhancement for multimodal image fusion with mamba. arXiv preprint arXiv:2404.09498, 2024a. Xinyu Xie, Xiaozhi Zhang, Xinglong Tang, Jiaxi Zhao, Dongping Xiong, Lijun Ouyang, Bin Yang, Hong Zhou, Bingo Wing-K...

arXiv 2023
[1988]

Adam: A method for stochastic optimization

Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980,

arXiv
[2014]

Zero-learning fast medical image fusion

Fayez Lahoud and Sabine S ¨usstrunk. Zero-learning fast medical image fusion. In 2019 22th Inter- national Conference on Information Fusion (FUSION), pp. 1–8. IEEE,

work page 2019
[2017]

Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challenge

Spyridon Bakas, Mauricio Reyes, Andras Jakab, Stefan Bauer, Markus Rempfler, Alessandro Crimi, Russell Takeshi Shinohara, Christoph Berger, Sung Min Ha, Martin Rozycki, et al. Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challenge. arXiv preprint arXiv:1...

arXiv

Show all 14 references

[2019]

Pyramid attention network for semantic segmentation

Hanchao Li, Pengfei Xiong, Jie An, and Lingxue Wang. Pyramid attention network for semantic segmentation. arXiv preprint arXiv:1805.10180,

arXiv
[2020]

Enhancing underwater imagery using gener- ative adversarial networks

Cameron Fabbri, Md Jahidul Islam, and Junaed Sattar. Enhancing underwater imagery using gener- ative adversarial networks. In 2018 IEEE International Conference on Robotics and Automation (ICRA), pp. 7159–7165. IEEE,

2018
[2021]

Mamba: Linear-time sequence modeling with selective state spaces

Albert Gu and Tri Dao. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752,

arXiv
[2022]

Mambadfuse: A mamba-based dual-phase model for multi-modality image fusion

Zhe Li, Haiwei Pan, Kejia Zhang, Yuhua Wang, and Fengming Yu. Mambadfuse: A mamba-based dual-phase model for multi-modality image fusion. arXiv preprint arXiv:2404.08406,

arXiv
[2023]

Very deep convolutional networks for large-scale image recognition

Karen Simonyan and Andrew Zisserman. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556,

arXiv
[2024]

An image is worth 16x16 words: Transformers for image recognition at scale

Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv...

2010 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [4]

Furthermore, compared to the FL1N method (Li et al., 2022), our approach produced brighter and more defined edges, thereby improving the contrast between edges and tissues

For the MRI-CT fusion task, our method demonstrated clear advantages over the FER strategy (Fu et al., 2021), achieving enhanced fidelity with sharper delineation of inner tissue struc- tures from MRI images and well-preserved boundaries from CT images. Furthermore, compared to the FL1N method (Li et al., 2022), our approach produced brighter and more def...

work page 2021

[2] [8]

Weisheng Li, Xiuxiu Peng, Jun Fu, Guofen Wang, Yuping Huang, and Feifei Chao

doi: 10.1109/TIP.2018.2887342. Weisheng Li, Xiuxiu Peng, Jun Fu, Guofen Wang, Yuping Huang, and Feifei Chao. A multiscale double-branch residual attention network for anatomical–functional medical image fusion. Com- puters in Biology and Medicine, 141:105005,

arXiv 2018

[3] [10]

The multimodal brain tumor image segmentation benchmark (brats)

Bjoern H Menze, Andras Jakab, Stefan Bauer, Jayashree Kalpathy-Cramer, Keyvan Farahani, Justin Kirby, Yuliya Burren, Nicole Porz, Johannes Slotboom, Roland Wiest, et al. The multimodal brain tumor image segmentation benchmark (brats). IEEE transactions on medical imaging , 34 (10):1993–2024,

1993

[4] [11]

Fusionmamba: Efficient image fusion with state space model

Siran Peng, Xiangyu Zhu, Haoyu Deng, Zhen Lei, and Liang-Jian Deng. Fusionmamba: Efficient image fusion with state space model. arXiv preprint arXiv:2404.07932,

arXiv

[5] [13]

Xinyu Xie, Yawen Cui, Chio-In Ieong, Tao Tan, Xiaozhi Zhang, Xubin Zheng, and Zitong Yu

doi: 10.1109/TIM.2023.3317470. Xinyu Xie, Yawen Cui, Chio-In Ieong, Tao Tan, Xiaozhi Zhang, Xubin Zheng, and Zitong Yu. Fusionmamba: Dynamic feature enhancement for multimodal image fusion with mamba. arXiv preprint arXiv:2404.09498, 2024a. Xinyu Xie, Xiaozhi Zhang, Xinglong Tang, Jiaxi Zhao, Dongping Xiong, Lijun Ouyang, Bin Yang, Hong Zhou, Bingo Wing-K...

arXiv 2023

[6] [1988]

Adam: A method for stochastic optimization

Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980,

arXiv

[7] [2014]

Zero-learning fast medical image fusion

Fayez Lahoud and Sabine S ¨usstrunk. Zero-learning fast medical image fusion. In 2019 22th Inter- national Conference on Information Fusion (FUSION), pp. 1–8. IEEE,

work page 2019

[8] [2017]

Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challenge

Spyridon Bakas, Mauricio Reyes, Andras Jakab, Stefan Bauer, Markus Rempfler, Alessandro Crimi, Russell Takeshi Shinohara, Christoph Berger, Sung Min Ha, Martin Rozycki, et al. Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challenge. arXiv preprint arXiv:1...

arXiv