Resource-Efficient Glioma Segmentation on Sub-Saharan MRI

Abbas Rabiu Muhammad; Abderrazek Zeraii; Brain Halubanza; Confidence Raymond; Dong Zhang; Freedmore Sidume; Hajer Ibrahim; Joseph Muthui Wacira; Olfa Gaddour; Oluwaseun Kalejaye

REVIEW 4 major objections 6 minor 17 references

Resource-Efficient Glioma Segmentation on Sub-Saharan MRI

T0 review · 4 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read A ~90 MB 3D Attention U-Net fine-tuned on Sub-Saharan MRI achieves Dice up to 0.85, showing lightweight models can handle low-resource glioma segmentation.

desk verdict A useful lightweight baseline for BraTS-Africa, but the evaluation section is too inconsistent to verify the headline Dice scores. read the letter →

arxiv 2509.09469 v1 pith:ZKEXMNGZ submitted 2025-09-11 cs.CV cs.AI

Freedmore Sidume , Oumayma Soula , Joseph Muthui Wacira , YunFei Zhu , Abbas Rabiu Muhammad , Abderrazek Zeraii , Oluwaseun Kalejaye , Hajer Ibrahim

show 5 more authors

Olfa Gaddour Brain Halubanza Dong Zhang Udunna C Anazodo Confidence Raymond

This is my paper

classification cs.CVcs.AI

keywords gliomasegmentationSub-SaharanAfricaMRI3DAttentionU-Nettransferlearninglow-resourcesettingsBraTS-AfricaresidualblocksTverskyloss

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

BrainUNet, a compact 3D Attention U-Net with residual blocks and attention gates, is fine-tuned on only 50 Sub-Saharan African MRI cases after pre-training on a large high-quality dataset. The paper reports that this lifts lesion subregion Dice scores from 0.52–0.62 to 0.76–0.85 on the BraTS-Africa benchmark, with a 91 MB model and sub-minute per-volume inference on a CPU. If those numbers are trustworthy, they would show that transfer learning plus a small local cohort is enough to bring automated glioma delineation into clinics that lack radiologists and GPU infrastructure. The paper's stated aim is to close the gap in equitable AI for global health, and its evidence suggests the main obstacle is domain shift rather than model size.

What carries the argument

The central mechanism is BrainUNet, a 3D U-Net with residual blocks and attention gates. Residual blocks—two 3D convolutions, batch normalization, ReLU, and a skip connection—stabilize training and preserve identity information. Attention gates on skip connections re-weight encoder features using a gating signal from the decoder, steering the network toward clinically relevant regions. The other load-bearing component is the two-stage training recipe: pre-training on 1,251 high-quality BraTS 2021 volumes followed by fine-tuning with a Tversky loss on BraTS-Africa. The paper presents this transfer-learning step as the difference between mediocre and clinically relevant scores.

What would settle it

Re-run the fine-tuned model on the BraTS-Africa cases with a strictly documented, case-disjoint split (e.g., a single 60/35 partition with no early stopping or hyperparameter selection based on the 35-case held-out set) and recompute Dice; if the scores fall well below 0.76/0.80/0.85, the generalization claim is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that domain adaptation via transfer learning—pre-training on 1,251 high-quality 3D scans, then fine-tuning on a small cohort of noisy, low-resolution Sub-Saharan scans—makes a lightweight attention-gated U-Net accurate enough for clinical glioma segmentation. Before fine-tuning the model scored 0.52–0.62 Dice across the three subregions; after fine-tuning the same architecture scores 0.76 (Enhancing Tumor), 0.80 (Necrotic/Non-Enhancing Core), and 0.85 (Surrounding FLAIR Hyperintensity) on the BraTS-Africa benchmark. The paper attributes the gain to residual blocks, attention gates that focus on salient regions, and augmentation that simulates motion and ghosting

Load-bearing premise

The reported Dice scores are treated as valid estimates of generalization, which presupposes that the fine-tuning training/validation split (stated as 50/10) and the 5-fold cross-validation partitions are properly disjoint and that no held-out case influenced model selection or early stopping.

Editorial extensions

If this is right

Fine-tuning a large pre-trained model on a small local dataset is a viable route to accurate segmentation where annotated data is scarce.
A ~91 MB, ~22M-parameter model can run a whole volume in 30–56 s on modest hardware, making deployment in non-GPU clinical settings plausible.
The large jump in Dice after fine-tuning implies that domain shift—not model capacity—is the primary barrier on Sub-Saharan MRI.
If the results generalize, the approach could support clinical decision-making in low-resource settings by providing automated tumor delineation without expert radiologists on site.
The model's whole-tumor Dice is competitive with heavier baselines, suggesting that resource-efficient architectures do not necessarily forfeit clinically useful accuracy.

Reading between the lines

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

Editorial inference: the same pre-train-then-fine-tune recipe may transfer to other lesion-segmentation tasks in low-resource settings (e.g., stroke or diabetic retinopathy), but the paper provides evidence only for glioma segmentation.
Editorial inference: if the split-integrity issue is resolved and the scores hold, the CPU runtime of ~56 s implies that even lower-end hardware could run the model, potentially opening the door to point-of-care use; the paper did not test this.
Editorial inference: the decision to drop the native T1 modality suggests that a three-modality acquisition might be sufficient, which could shorten scan times; the paper does not analyze this trade-off.
Editorial inference: a practical falsifiable extension would be to measure how Dice changes as the fine-tuning cohort shrinks (e.g., 10, 20, 50 cases) to identify the minimum viable annotation budget.

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

First, the bottom line: this is a practical application note—a compact 3D Attention U-Net with residual blocks, pre-trained on BraTS 2021 and fine-tuned on BraTS-Africa, about 91 MB, sub-minute CPU inference. That combination is genuinely relevant for low-resource settings, and the authors are honest about the limitations.

The paper does a few things well. The two-stage transfer learning strategy is sensible, the runtime comparison on CPU/GPU is useful, and the comparison to nnUNet and MedNeXt gives context. The limitations section is candid about data size, image quality, and single-dataset evaluation.

The problem is the evaluation section. Section 2.1 says 60 training/35 validation cases; Section 2.4 says fine-tuning used 50 training/10 validation; the 'Evaluation Metrics' paragraph says 5-fold cross-validation with validation n=35 per fold, which is arithmetically impossible for 95 cases. It is never stated whether the Synapse benchmark was a held-out set, or whether the 10 validation cases were used for model selection. Table 2 reports Dice 0.76/0.80/0.85, but Table 3 gives BrainUNet a lesion-wise average of 0.743 and legacy average of 0.806—two different sets of numbers for what should be the same model. Figure 5's caption says 'before fine-tuning' while the text says it shows improvement after fine-tuning. These are not minor typos; they make it impossible to verify the abstract's headline numbers.

The underlying finding—that fine-tuning helps—is plausible and consistent with prior work [3,5]. The exact gains are unverified. The paper also doesn't release code with a usable commit; the GitHub link points to a directory.

Who is this for? Researchers working on low-resource MRI segmentation, especially for African populations. It's not a methodological advance, but a useful engineering baseline if the evaluation is straightened out. I would send it to peer review—the dataset is important and the deployment message is valuable—but I would require a clear split description, a held-out test set, error bars, and code release before accepting. As written, the headline Dice are not trustworthy.

Referee Report

4 major / 6 minor

Summary. The paper proposes BrainUNet, a 3D Attention U-Net with residual blocks and attention gates, pre-trained on BraTS 2021 and fine-tuned on the BraTS-Africa dataset (95 MRI cases). The central claim is that fine-tuning yields Dice scores of 0.76 (ET), 0.80 (NETC), 0.85 (SNFH) while remaining lightweight (~91 MB, sub-minute inference on CPU/GPU), demonstrating that a compact model can generalize to Sub-Saharan African MRI. The paper also reports a comparison with nnUNet and MedNeXt on BraTS-Africa and discusses deployment feasibility for low-resource settings.

Significance. If the reported Dice scores are valid estimates of held-out generalization, the contribution is practically meaningful: a low-footprint model that maintains competitive segmentation accuracy on a difficult, low-resource MRI dataset would be a useful step toward equitable AI in global health. The authors are explicit about the model's compact size, inference time, and the use of transfer learning, and they state limitations (small dataset, single-site external validation). However, the paper's evaluation protocol is described inconsistently, and the headline numbers are not tied to a clearly disjoint train/validation/test partition. Because the main claim is empirical, the credibility of the exact Dice scores depends on resolving these protocol issues.

major comments (4)

[§2.1, §2.4 (Evaluation Metrics), Table 2] The split descriptions are mutually contradictory. §2.1 says BraTS-Africa has 60 training / 35 validation cases; §2.4 says fine-tuning used 50 training / 10 validation samples; the Evaluation Metrics paragraph says 5-fold cross-validation with one partition as validation (n=35). With 95 cases, a 5-fold partition would be 19 cases, not 35, so the text conflates at least two different protocols. The paper never states whether the 10 validation cases are disjoint from the 35 official validation cases, whether the 35 were used for model selection, or whether the Synapse benchmark cases were a held-out set. If the hyperparameters or early stopping used the same cases later reported in Table 2, the Dice scores are optimistic estimates of generalization. This is the load-bearing issue for the paper's central claim.
[Table 2 vs. Table 3] Table 2 reports BrainUNet fine-tuned Dice of 0.76/0.80/0.85 (ET/NETC/SNFH), while Table 3 reports BrainUNet lesion-wise Dice of 0.684/0.714/0.831 and legacy Dice of 0.759/0.791/0.869, with averages 0.743 and 0.806. These do not match Table 2. The reader cannot tell which split, which preprocessing, or which evaluation mode produced the headline numbers. The authors should report one consistent evaluation protocol and state exactly which table corresponds to which partition and whether Synapse benchmarking was on a held-out set.
[Figure 5 caption and Figures 5–6] Figure 5's caption says 'Segmentation results with BrainUNet before fine-tuning,' but the surrounding text and Table 2 emphasize that fine-tuning markedly improves segmentation. If Figure 5 is intended to show the improvement, the caption is wrong; if it is intentionally before-only, it does not support the fine-tuning claim. Figure 6's caption says the model was fine-tuned on 'BraTS-Africa validation data,' which is troubling: fine-tuning on validation data and then reporting that data as validation is circular. The authors should clarify which cases were used for fine-tuning, validation, and final evaluation.
[§3.1 and §3.2, Dice variability] The 5-fold cross-validation section reports average validation Dice stabilizing at 0.55 before fine-tuning, while the fine-tuned Synapse evaluation reports 0.76–0.85. No error bars, per-fold ranges, or statistical tests are given. With only 10 validation samples in the fine-tuning split, the reported differences could be within chance variation. The authors should report per-fold scores and confidence intervals, and clearly separate pre-fine-tuning CV results from post-fine-tuning held-out results.

minor comments (6)

[Abstract and §2.1] The abstract uses 'Surrounding Non-Functional Hemisphere' as the expansion of SNFH; the correct BraTS term is 'Surrounding Non-Enhancing FLAIR Hyperintensity' (used in §2.1). Please fix this terminology for consistency and correctness.
[Table 1] The Tversky loss weights α and β are not reported, although they are free parameters in Eq. (1). Reporting their values is necessary for reproducibility.
[§2.4, Eq. (1)] The notation in the Tversky loss equation is compressed; please define the sums explicitly (over all voxels and classes) and ensure the equation matches the implementation.
[Section 2.2] Section 2.2 is empty ('2.2 Proposed Approach') and is immediately followed by '2.3 Proposed BrainUNet Framework.' This is a formatting error that should be corrected.
[Fig. 5] The representative case in Figure 5 is not identified. Providing a case ID would help the reader connect the qualitative result to the quantitative tables, especially given the caption inconsistency.
[Reference [8]] Reference [8] is cited for the Adam optimizer, but the cited paper concerns RMSProp. Please cite the original Adam paper or adjust the text to match the reference.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; minor self-citation only, and evaluation-protocol ambiguities are correctness concerns rather than circular reasoning.

full rationale

The paper is an empirical deep-learning study: it proposes a 3D Attention U-Net with residual blocks, pre-trains on BraTS 2021, fine-tunes on BraTS-Africa, and reports Dice/Hausdorff metrics on a validation set benchmarked via the Synapse platform. There is no mathematical derivation chain in which a predicted quantity is defined in terms of the input or in which a fitted parameter is renamed as a prediction. The only self-citation is reference [17] (Zhang et al.), used as a footnote link to the BrainUNet repository; it is not load-bearing for the paper's quantitative claims, and the architecture is described in full in Section 2.4. The manuscript does, however, contain internal inconsistencies in the reported split sizes (60/35 in Section 2.1, 50/10 in Section 2.4, and a 5-fold CV with an n=35 validation partition), and Figure 6's caption says the model was 'fine-tuned on BraTS-Africa validation data,' which could be misread as training on the validation set. These issues bear on whether the reported Dice scores are unbiased estimates of generalization, but they are evaluation-protocol concerns, not circular reasoning. The paper also states its own limitations (small dataset, single external validation) in Section 4. Under the definitions of circularity used here (self-definition, fitted-input-as-prediction, load-bearing self-citation, imported uniqueness, ansatz-by-citation, renaming), no step in the paper reduces to its own input. The modest score of 2 reflects only the presence of a minor, non-load-bearing self-citation and the unresolved ambiguity around validation data, not a circular derivation.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central numbers are empirical results of a trained model, so the underlying axioms are domain assumptions about the data and the transfer learning premise. The free parameters are hyperparameters and preprocessing choices that are either unreported or chosen without ablation.

free parameters (4)

Tversky loss weights alpha and beta
These balance false positives and false negatives in Eq. (1); their values are not reported, yet they directly affect the trained model and the reported Dice scores.
Fine-tuning learning rate = 1e-4
Chosen by hand (Table 1); influences convergence and final Dice.
Fine-tuning epochs = 50
Chosen by hand (Table 1); no stopping criteria or early stopping reported.
Input crop size = 128x128x128
Chosen to reduce compute (Section 2.3); no ablation on cropping size.

assumptions (3)

domain assumption BraTS-Africa labels are accurate expert annotations
The model is trained and evaluated against these labels (Section 2.1); if labels are noisy, Dice scores are biased.
domain assumption Pre-training on BraTS 2021 transfers to SSA MRI
The two-stage training relies on this premise; the paper only shows before/after fine-tuning on the same dataset, not an independent test of transfer.
ad hoc to paper Cropping to 128x128x128 voxels preserves tumor-relevant information
Chosen for efficiency without experimentation (Section 2.3); large tumors or off-center anatomy could be truncated.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Resource-Efficient Glioma Segmentation on Sub-Saharan MRI." pith.science (2026). https://pith.science/paper/ZKEXMNGZ

@misc{pith2026250909469,
  author       = {Pith},
  title        = {Pith review of: Resource-Efficient Glioma Segmentation on Sub-Saharan MRI},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZKEXMNGZ}},
  note         = {Machine review of arXiv:2509.09469}
}

read the original abstract

Gliomas are the most prevalent type of primary brain tumors, and their accurate segmentation from MRI is critical for diagnosis, treatment planning, and longitudinal monitoring. However, the scarcity of high-quality annotated imaging data in Sub-Saharan Africa (SSA) poses a significant challenge for deploying advanced segmentation models in clinical workflows. This study introduces a robust and computationally efficient deep learning framework tailored for resource-constrained settings. We leveraged a 3D Attention UNet architecture augmented with residual blocks and enhanced through transfer learning from pre-trained weights on the BraTS 2021 dataset. Our model was evaluated on 95 MRI cases from the BraTS-Africa dataset, a benchmark for glioma segmentation in SSA MRI data. Despite the limited data quality and quantity, our approach achieved Dice scores of 0.76 for the Enhancing Tumor (ET), 0.80 for Necrotic and Non-Enhancing Tumor Core (NETC), and 0.85 for Surrounding Non-Functional Hemisphere (SNFH). These results demonstrate the generalizability of the proposed model and its potential to support clinical decision making in low-resource settings. The compact architecture, approximately 90 MB, and sub-minute per-volume inference time on consumer-grade hardware further underscore its practicality for deployment in SSA health systems. This work contributes toward closing the gap in equitable AI for global health by empowering underserved regions with high-performing and accessible medical imaging solutions.

Figures

Figures reproduced from arXiv: 2509.09469 by the authors.

**Figure 2.** 3D Attention U-Net with residual connections. Data Preprocessing MRI scans were cropped to 128 × 128 × 128 to reduce computational demand. Three modalities (FLAIR, T1CE, T2W) were stacked, providing optimal tumor contrast [1]. Percentile clipping mitigated outliers, while intensity normalization harmonized modalities and masks [12]. These steps improved efficiency and preserved critical information for robust segme… view at source ↗

**Figure 3.** Core components of BrainUNet: (a) residual block with skip connections, and (b) attention block with gating for region-specific focus. Transfer Learning: Before fine-tuning on BraTS-Africa, the model was pretrained on the larger BraTS-GLI 2021 dataset (n = 1251 3D MRI scans). Training used the parameters in [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Average training and validation metrics across folds during cross-validation. (Left) Average Dice Score, and (Right) Average Loss. These figures reflect the performance of the BrainUNet model before fine-tuning. 3.2 Synapse Evaluation: BrainUNet Before and After Fine-…

**Figure 5.** Figure 5: Segmentation results with BrainUNet before fine-tuning. T1CE, T2W, and FLAIR modalities are shown with ground-truth (Mask) and predicted segmentation (Pred). Model Lesion-wise Dice Avg Legacy Dice Avg Params ET TC WT ET TC WT nnUNet 0.797 0.786 0.846 0.810 0.850 0.853 …

**Figure 6.** Figure 6: Final segmentation results from BrainUNet fine-tuned on BraTS-Africa validation data [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

17 extracted references · 8 canonical work pages

[1]

Abidin, Z.U., Naqvi, e.a.: Recent deep learning-based brain tu- mor segmentation models using multi-modality magnetic reso- nance imaging: a prospective survey. Front. Bioeng. Biotech- nol.12(Jul 2024). https://doi.org/10.3389/fbioe.2024.1392807, https://www.frontiersin.org/journals/bioengineering-and- biotechnology/articles/10.3389/fbioe.2024.1392807/ful...

arXiv 2024
[2]

https://doi.org/10.48550/arXiv.2305.19369, http://arxiv.org/abs/2305.19369, arXiv:2305.19369 [physics]

Adewole, M., Rudie, e.a.: The Brain Tumor Segmentation (BraTS) Chal- lenge 2023: Glioma Segmentation in Sub-Saharan Africa Patient Popula- tion (BraTS-Africa) (May 2023). https://doi.org/10.48550/arXiv.2305.19369, http://arxiv.org/abs/2305.19369, arXiv:2305.19369 [physics]

arXiv doi:10.48550/arxiv.2305.19369 2023
[3]

Bridging the Gap: Generalising State-of-the-Art U-Net Models to Sub-Saharan African Populations

Amod, A.R., Smith, e.a.: Bridging the Gap: Generalising State-of-the- Art U-Net Models to Sub-Saharan African Populations (Dec 2023). https://doi.org/10.48550/arXiv.2312.11770, http://arxiv.org/abs/2312.11770, arXiv:2312.11770 [cs, eess]

work page Pith review arXiv doi:10.48550/arxiv.2312.11770 2023
[4]

https://doi.org/10.48550/arXiv.2107.02314, http://arxiv.org/abs/2107.02314, arXiv:2107.02314 [cs]

Baid, U., Ghodasara, e.a.: The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification (Sep 2021). https://doi.org/10.48550/arXiv.2107.02314, http://arxiv.org/abs/2107.02314, arXiv:2107.02314 [cs]

arXiv doi:10.48550/arxiv.2107.02314 2021
[5]

Towards SAMBA: Segment Anything Model for Brain Tumor Segmentation in Sub-Sharan African Populations

Barakat, M., Magdy, e.a.: Towards SAMBA: Segment Anything Model for Brain Tumor Segmentation in Sub-Sharan African Populations (Dec 2023). https://doi.org/10.48550/arXiv.2312.11775, http://arxiv.org/abs/2312.11775, arXiv:2312.11775 [cs, eess] Resource-Efficient Glioma Segmentation on Sub-Saharan MRI 11

work page Pith review arXiv doi:10.48550/arxiv.2312.11775 2023
[6]

Visual Attention Methods in Deep Learning: An In-Depth Survey

Hassanin, M., Anwar, S., Radwan, I., Khan, F.S., Mian, A.: Visual At- tention Methods in Deep Learning: An In-Depth Survey (May 2024). https://doi.org/10.48550/arXiv.2204.07756, http://arxiv.org/abs/2204.07756, arXiv:2204.07756 [cs, eess]

work page Pith review arXiv doi:10.48550/arxiv.2204.07756 2024
[7]

Medical Image Analysis36, 61–78 (Feb 2017)

Kamnitsas, K., Ledig, e.a.: Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation. Medical Image Analysis36, 61–78 (Feb 2017). https://doi.org/10.1016/j.media.2016.10.004, https://www.sciencedirect.com/science/article/pii/S1361841516301839

doi:10.1016/j.media.2016.10.004 2017
[8]

Training of Deep Neural Networks based on Distance Measures using RMSProp

Kurbiel, T., Khaleghian, S.: Training of Deep Neural Net- works based on Distance Measures using RMSProp (Aug 2017). https://doi.org/10.48550/arXiv.1708.01911, http://arxiv.org/abs/1708.01911, arXiv:1708.01911 [cs, stat]

work page Pith review arXiv doi:10.48550/arxiv.1708.01911 2017

Show all 17 references

[9]

NMR in Biomedicine37(7), e4992 (Jul 2024)

Murali, S., Ding, H., Adedeji, F., Qin, C., Obungoloch, J., Asllani, I., Anazodo, U., Ntusi, N.A.B., Mammen, R., Niendorf, T., Adeleke, S.: Bringing MRI to low- and middle-income countries: directions, challenges and potential solutions. NMR in Biomedicine37(7), e4992 (Jul 202...

2024 doi
[10]

J Digit Imag- ing33(4), 903–915 (Aug 2020)

Mzoughi, H., Njeh, e.a.: Deep Multi-Scale 3D Convolutional Neural Net- work (CNN) for MRI Gliomas Brain Tumor Classification. J Digit Imag- ing33(4), 903–915 (Aug 2020). https://doi.org/10.1007/s10278-020-00347-9, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7522155/

2020 doi
[11]

BMC Health Serv Res23, 787 (Jul 2023)

Nigatu, A.M., Yilma, T.M., Gezie, L.D., Gebrewold, Y., Gullslett, M.K., Mengiste, S.A., Tilahun, B.: Medical imaging consultation practices and challenges at public hospitals in the Amhara regional state, North- west Ethiopia: a descriptive phenomenological study. BMC Health S...

2023 doi
[12]

Sci Rep11, 10930 (May 2021)

Ranjbarzadeh, R., Bagherian Kasgari, e.a.: Brain tumor segmentation based on deep learning and an attention mechanism using MRI multi-modalities brain im- ages. Sci Rep11, 10930 (May 2021). https://doi.org/10.1038/s41598-021-90428-8, https://www.ncbi.nlm.nih.gov/pmc/articles/P...

2021 doi
[13]

https://doi.org/10.48550/arXiv.2401.01303, http://arxiv.org/abs/2401.01303, arXiv:2401.01303 [cs, eess]

Sahayam, S., Jayaraman, U.: Integrating Edges into U-Net Models with Explain- able Activation Maps for Brain Tumor Segmentation using MR Images (Jan 2024). https://doi.org/10.48550/arXiv.2401.01303, http://arxiv.org/abs/2401.01303, arXiv:2401.01303 [cs, eess]

2024 doi arXiv
[14]

Journal of Computational Sci- ence30, 174–182 (Jan 2019)

Sajjad, M., Khan, e.a.: Multi-grade brain tumor classification using deep CNN with extensive data augmentation. Journal of Computational Sci- ence30, 174–182 (Jan 2019). https://doi.org/10.1016/j.jocs.2018.12.003, https://www.sciencedirect.com/science/article/pii/S1877750318307385

2019 doi
[15]

https://doi.org/10.48550/arXiv.1706.05721, http://arxiv.org/abs/1706.05721, arXiv:1706.05721 [cs]

Salehi, S.S.M., Erdogmus, D., Gholipour, A.: Tversky loss function for im- age segmentation using 3D fully convolutional deep networks (Jun 2017). https://doi.org/10.48550/arXiv.1706.05721, http://arxiv.org/abs/1706.05721, arXiv:1706.05721 [cs]

2017 doi arXiv
[16]

Tajbakhsh, N., Shin, e.a.: Convolutional Neural Networks for Medical Im- age Analysis: Full Training or Fine Tuning? IEEE Trans. Med. Imag- ing35(5), 1299–1312 (May 2016). https://doi.org/10.1109/TMI.2016.2535302, http://arxiv.org/abs/1706.00712, arXiv:1706.00712 [cs]

2016
[17]

In: International Conference on Medical Image Computing and Computer-Assisted Intervention

Zhang, D., Confidence, R., Anazodo, U.: Stroke lesion segmentation from low- quality and few-shot mris via similarity-weighted self-ensembling framework. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 87–96. Springer (2022)

2022

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

Abidin, Z.U., Naqvi, e.a.: Recent deep learning-based brain tu- mor segmentation models using multi-modality magnetic reso- nance imaging: a prospective survey. Front. Bioeng. Biotech- nol.12(Jul 2024). https://doi.org/10.3389/fbioe.2024.1392807, https://www.frontiersin.org/journals/bioengineering-and- biotechnology/articles/10.3389/fbioe.2024.1392807/ful...

arXiv 2024

[2] [2]

https://doi.org/10.48550/arXiv.2305.19369, http://arxiv.org/abs/2305.19369, arXiv:2305.19369 [physics]

Adewole, M., Rudie, e.a.: The Brain Tumor Segmentation (BraTS) Chal- lenge 2023: Glioma Segmentation in Sub-Saharan Africa Patient Popula- tion (BraTS-Africa) (May 2023). https://doi.org/10.48550/arXiv.2305.19369, http://arxiv.org/abs/2305.19369, arXiv:2305.19369 [physics]

arXiv doi:10.48550/arxiv.2305.19369 2023

[3] [3]

Bridging the Gap: Generalising State-of-the-Art U-Net Models to Sub-Saharan African Populations

Amod, A.R., Smith, e.a.: Bridging the Gap: Generalising State-of-the- Art U-Net Models to Sub-Saharan African Populations (Dec 2023). https://doi.org/10.48550/arXiv.2312.11770, http://arxiv.org/abs/2312.11770, arXiv:2312.11770 [cs, eess]

work page Pith review arXiv doi:10.48550/arxiv.2312.11770 2023

[4] [4]

https://doi.org/10.48550/arXiv.2107.02314, http://arxiv.org/abs/2107.02314, arXiv:2107.02314 [cs]

Baid, U., Ghodasara, e.a.: The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification (Sep 2021). https://doi.org/10.48550/arXiv.2107.02314, http://arxiv.org/abs/2107.02314, arXiv:2107.02314 [cs]

arXiv doi:10.48550/arxiv.2107.02314 2021

[5] [5]

Towards SAMBA: Segment Anything Model for Brain Tumor Segmentation in Sub-Sharan African Populations

Barakat, M., Magdy, e.a.: Towards SAMBA: Segment Anything Model for Brain Tumor Segmentation in Sub-Sharan African Populations (Dec 2023). https://doi.org/10.48550/arXiv.2312.11775, http://arxiv.org/abs/2312.11775, arXiv:2312.11775 [cs, eess] Resource-Efficient Glioma Segmentation on Sub-Saharan MRI 11

work page Pith review arXiv doi:10.48550/arxiv.2312.11775 2023

[6] [6]

Visual Attention Methods in Deep Learning: An In-Depth Survey

Hassanin, M., Anwar, S., Radwan, I., Khan, F.S., Mian, A.: Visual At- tention Methods in Deep Learning: An In-Depth Survey (May 2024). https://doi.org/10.48550/arXiv.2204.07756, http://arxiv.org/abs/2204.07756, arXiv:2204.07756 [cs, eess]

work page Pith review arXiv doi:10.48550/arxiv.2204.07756 2024

[7] [7]

Medical Image Analysis36, 61–78 (Feb 2017)

Kamnitsas, K., Ledig, e.a.: Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation. Medical Image Analysis36, 61–78 (Feb 2017). https://doi.org/10.1016/j.media.2016.10.004, https://www.sciencedirect.com/science/article/pii/S1361841516301839

doi:10.1016/j.media.2016.10.004 2017

[8] [8]

Training of Deep Neural Networks based on Distance Measures using RMSProp

Kurbiel, T., Khaleghian, S.: Training of Deep Neural Net- works based on Distance Measures using RMSProp (Aug 2017). https://doi.org/10.48550/arXiv.1708.01911, http://arxiv.org/abs/1708.01911, arXiv:1708.01911 [cs, stat]

work page Pith review arXiv doi:10.48550/arxiv.1708.01911 2017