Jul 2026· 2026 5th International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE)· pp. 1-8· 0 citations· 21 references
Abstract
In order to make accurate distinctions within gliomas - specifically necrosis, edema, and enhancement - accurate segmentation of these regions from multiple MRI types is required. This task has proven difficult because of how heavily class imbalance affects voxels in this imaging modality, combined with its high degree of morphologic variability. In response to this challenge, a new compact 3D attention-based U-Net is developed that uses multi-task regularization, which is used for the volumetric segmentation of gliomas using data from the BraTS 2023 competition. By applying attention gates at each skip connection in the proposed model, non-tumor-related activations are suppressed before the output of the decoder is concatenated together. Additionally, the addition of a lightweight binary auxiliary classification layer after the first bottleneck in the model can act as an inductive bias to prevent the encoder from collapsing into shortcut texture features based solely on local information when faced with severe class imbalance. The model was tested on a held-out set of 251 patients, resulting in Dice similarity coefficients of 0.927 (whole tumor), 0.877 (tumor core), and 0.831 (enhancing tumor). Furthermore, the median HD95 found for whole tumor boundary localization was 1.57 mm. The model has approximately 5.64 million trainable parameters and performs inference in approximately 0.18 s per patient on a single GPU. This represents a 70% decrease in the number of trainable parameters compared to standard 3D U-Net architectures performing similarly well on segmentation tasks, thus making it suitable for use in resource-constrained deployment settings.
The problem of accurate identification of brain tumors using multi-modal MRIs still poses significant challenges due to tumor heterogeneity and variations across different image modes, as well as inconsistent availability of imaging modalities in real-world applications. Although deep learning algorithms such as convolution and transformer networks have demonstrated high efficacy in tumor segmentation through effective modeling of local and global contextual information, many state-of-the-art models perform directly on concatenated or fused images, making it difficult to leverage differences in information provided by different modalities and leading to inferior performance in cases of modality imbalance or missing data. To tackle the issue, we present in this paper a novel multimodal transformer network using the concept of reliability-driven modality attention for robust brain tumor segmentation. Our approach employs a feature extraction pipeline with a reliability estimator that automatically calculates weighting coefficients for each input modality (T1, T2, FLAIR, T1-CE), enabling more efficient feature representation than traditional fusion techniquesFurthermore, a slice-aware 2.5D context modeling strategy is used to capture inter-slice dependencies while keeping computational efficiency high compared to full 3D models. Extensive experiments on benchmark multi-modal MRI datasets show that the proposed approach achieves better segmentation performance than leading CNN, transformer, and hybrid methods, especially in scenarios with missing or degraded modalities. The results emphasize how reliability-aware fusion improves robustness, generalization, and clinical use of automated brain tumor analysis systems.
C. Nirmala, T. R. Ganesh Babu· International Conference on...· 0 citations
This study proposes a fully unsupervised brain tumor segmentation framework using multimodal MRI data that enables precise tumor delineation and is suitable for large-scale clinical integration.
J. Adlin, Arockia Selva Saroja· ITEGAM- Journal of Engineeri...· 0 citations
Brain tumor segmentation from Magnetic Resonance Imaging (MRI) is an important task in computer-aided diagnosis because accurate identification of tumor regions supports clinical assessment and treatment planning. However, the complex structure, irregular shape, intensity variation, and heterogeneous appearance of brain tumors make automated segmentation challenging. This study presents a comparative deep learning framework for brain tumor segmentation using the BraTS 2020 dataset and two-dimensional (2D) MRI images. In this study, we explore 2D deep learning architectures for automated tumor segmentation, focusing on U-Net, Vision Transformer (ViT), and Res-ViT models. U-Net, with its encoder–decoder design and skip connections, has been widely adopted for medical image segmentation due to its ability to capture fine-grained spatial features. ViT, leveraging self-attention mechanisms, introduces a global receptive field that enhances contextual understanding across slices. The Res-ViT hybrid combines residual learning with transformer-based attention, aiming to balance local feature extraction and long-range dependency modelling. Preprocessing steps, including skull stripping, intensity normalisation, and bias field correction, were applied to ensure consistency across scans. Data augmentation techniques such as rotation, flipping, and elastic deformation were employed to mitigate overfitting and improve generalisation. The models are evaluated using important segmentation metrics, including Intersection over Union (IoU), accuracy, precision, recall/sensitivity, loss, and 95th-percentile Hausdorff Distance (HD95). The comparative analysis aims to identify the strengths and limitations of convolutional and transformer-based approaches for 2D brain tumor segmentation. The study demonstrates the potential of combining local feature extraction and global contextual learning to achieve more accurate and robust brain tumor segmentation from multimodal MRI images.
Keywords: MRI; U-Net; ViT; BraTS 2020; IOU; HD95.
Lovedeep Kaur, Parminder Singh, Naveen Dhillon· International Journal of Com...· 0 citations
Brain tumor segmentation from MRI is clinically critical yet challenging due to heterogeneous appearance and irregular boundaries. Conventional CNN based methods lack effective global context modeling, while transformer-based approaches are computationally expensive and unstable on limited datasets. To address these, we propose MMA-UNet, a novel hybrid architecture that integrates multiple complementary mechanisms within an encoder-decoder framework. The model employs an EfficientNet-B5 encoder, MedNeXt bridge blocks, a Mamba-based VSS bottleneck, a CBAM enhanced decoder, and deformable refinement for precise boundary adaptation. The model achieves a Dice score of 0.9063, IoU of 0.8304, precision of 0.9043, recall of 0.9092, and specificity of 0.9982 on FigShare benchmark, outperforming the evaluated U-Net, Attention U-Net, TransUNet and Swin UNet under the adopted experimental protocol and maintaining parameter efficiency (30.41 M), demonstrating strong robustness on T1-weighted contrast-enhanced MRI. The proposed model operates on independent 2D slices without volumetric context, and its generalizability to larger datasets remains to be established.
Manual delineation of ischemic stroke lesions on brain MRI is time-consuming and subject to disagreement between readers, which motivates reproducible automated methods. This work develops an Attention-Gated 3D U-Net for lesion delineation from three MRI sequences, namely diffusion-weighted imaging (DWI), the apparent diffusion coefficient (ADC) map, and fluid-attenuated inversion recovery (FLAIR). The study uses 250 subjects, each accompanied by an expert-drawn binary lesion mask. A four-level encoder-decoder with attention gates on its skip connections is trained under a joint Dice and binary crossentropy objective. Prior to training, every volume is resampled to 1.0 mm isotropic spacing, z-score normalized over in-brain voxels, and cropped to the brain region. Augmentation is applied during training and combines spatial and intensity perturbations. Optimization proceeds for 116 epochs using AdamW with a cosineannealed learning rate, attaining a best validation Dice of 0.6821. Evaluation employs both five-fold cross-validation and a held-out partition, while mixed-precision training, gradient accumulation, and a channels-last memory layout keep memory consumption modest. The attention gates concentrate the network response on lesion tissue and reduce spurious activations elsewhere.
H. V, A. V.· 2026 International Conferenc...· 0 citations
Background Accurate segmentation of glioma subregions from multimodal magnetic resonance imaging (MRI) is essential for diagnosis, treatment planning, and response assessment, but remains challenging because of boundary ambiguity, heterogeneous appearance, and small enhancing tumor (ET) components. This study aimed to develop and evaluate a controlled nnU-Net v2-based framework for three-dimensional (3D) brain tumor segmentation (BraTS) by improving adaptive feature representation and boundary-aware learning while preserving the reproducibility of the self-configuring nnU-Net pipeline. Methods We propose a conditional convolution and squeeze-and-excitation with boundary-aware learning network (CondSEB-Net), a controlled enhancement of the nnU-Net v2 framework for 3D medical image segmentation. Conditional convolution (CondConv) is introduced to improve sample-specific feature adaptation, multi-level 3D squeeze-and-excitation (SE) attention is used to recalibrate channel responses, and a boundary-aware loss based on signed distance maps (SDMs) is incorporated to strengthen contour-level supervision. The proposed method was evaluated on BraTS2020 and kidney tumor segmentation (KiTS)2019 using fixed five-fold cross-validation and compared with representative 3D segmentation baselines. Results On BraTS2020, CondSEB-Net improved the average Dice score from 0.8525±0.0217 to 0.8659±0.0184 and reduced the average 95% Hausdorff distance (HD95) from 10.57±4.01 to 9.08±3.64 mm compared with nnU-Net v2. The average symmetric surface distance (ASSD) also decreased from 3.08±0.37 to 2.65±0.30 mm. Fold-wise paired statistical analysis showed statistically supported improvements in both region overlap and boundary-related metrics. On KiTS2019, CondSEB-Net also achieved superior or competitive performance with limited additional computational overhead. Conclusions CondSEB-Net improves adaptive feature representation and boundary-related segmentation performance while preserving the self-configuring pipeline of nnU-Net v2. The current results provide benchmark-level evidence for the proposed framework, while external multi-center validation remains necessary before clinical deployment.
Cheng-Hong Zhang, Qiang Wei· Quantitative Imaging in Medi...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
