Jul 2026· IEEE International Conference on Circuits and Systems for Communications· pp. 1-6· 0 citations· 19 references
Abstract
With all its innovations, artificial intelligence (AI) has enabled unprecedented breakthroughs in many sectors such as medical image analysis, which may involve diagnosis, and treatment planning. Segmentation of medical images is an essential operation within this framework since it makes possible the identification of organs, tumors, and pathological regions within the image. However, although this is one of the most important aspects of medical image processing, segmentation still continues to pose some challenges. Traditional segmentation techniques struggle to segment complex medical images, especially in the presence of noise, intensity variations, and unclear boundaries. For this reason, Deep learning becomes the primary solution to overcome the limitations that hinder segmentation performance. In this overview, recently proposed supervised deep learning methods of medical image segmentation, including CNN-based, Transformer-based, and hybrid models, focusing on selected works related to breast and brain tumor segmentation and compares them according to model family, and reported performance. The paper also discusses current challenges and future research directions for developing more reliable and clinically useful segmentation models.
The most important step in diagnosis, treatment planning, and prediction analysis is the accurate segmentation of brain tumours from magnetic resonance imaging (MRI) scans. Radiologists' manual segmentation was time-consuming, subjective, and inaccurate. For this reason, there is a need for automated approaches. In Current days, deep learning (DL) has shown great promise for enhancing the processing of medical images. In DL, the U-Net architecture has become a typical framework for medical image segmentation of images due to its symmetric encoder–decoder design and skip links that preserve spatial detail. At the same time, conventional U-Net models are restricted to 2D slices and cannot detect contextual connections between slices in spatial MRI images; this may lead to discontinuities and reduced accuracy. The study suggests addressing the above-mentioned challenges, a hybrid deep learning framework consisting of a U-Net architecture integrated with Long Short-Term Memory (LSTM) networks is designed. The U-Net component extracts the variant-invariant spatial and structural features, and LSTM is responsible for integrating the temporal dependencies spatially, which inject discontinuities across parallel slices, which is imperative for the boundary delineation and localisation of the tumour. Compared to the baseline model, the suggested hybrid U-Net and LSTM networks exhibit noticeably better segmentation accuracy and visual consistency under various scenarios after being trained and assessed on a publicly accessible brain MRI dataset. According to experimental data, the suggested model provides great segmentation performance, with Dice scores above 95% and accuracy above 96%. To sum up, the whole exercise displays the promise of combining a convolutional and a recurrent architecture to propagate automated neuroimaging analysis. This will not only reduce the manual labour but also continue to work well in clinical practice, as it is comfortable and prepared for replication in the future.
K. Pranitha, Vuda Sreenivasa Rao· Pertanika journal of science...· 0 citations
Abstract Objectives To address the challenges of complex feature variations and unclear boundary definitions between segmented targets and surrounding regions in medical images, a novel segmentation model based on Deformable Large Kernel Convolutional Attention (D-LKA) and Transformer is proposed. Methods The model first uses Vision Transformer as the encoder to enhance the ability to capture global information, overcoming the limitations of convolutional neural networks’ receptive field. In the decoder, a D-LKA decoder with deformable large kernel convolution attention is used, allowing the model to adapt to complex target features. Finally, the TRR module is introduced to coordinate information transfer between the convolutional neural network and Transformer, reducing semantic loss. Results The model is trained, validated, and tested on the Kvasir-SEG colon polyp dataset, with multiple ablation experiments. To validate generalization, experiments are also conducted on the CVC-ClinicDB dataset. Experimental results show that TDU-Net outperforms other methods in both segmentation accuracy and generalization. Conclusions TDU-Net achieves excellent segmentation results, addressing small target and incomplete feature extraction issues, significantly improving clinical diagnosis efficiency and accuracy.
Segmentation of medical images is a crucial process for diagnosis and treatment planning. Yet, traditional CNN based models are often not able to convey high-level spatial relationships and intricate tissue boundaries in high-resolution medical images. To tackle these issues, this paper presents a novel ResUNet++–Vision Transformer (ResUNet++–ViT) framework incorporating both multi-scale local feature extraction and global contextual learning. The ResUNet++ backbone consists of residual blocks and nested skip connections, which are used to extract hierarchical features, and the Vision Transformer uses self-attention mechanisms to capture long-range dependencies. A fusion module allows for the integration of local and global features, resulting in better segmentation accuracy and preserving the boundary. The proposed model was tested on the ISBI 2012 Electron Microscopy Segmentation Challenge (EMSC) dataset, and obtained a Dice score of 0.960, IoU of 0.920, precision of 0.967, recall of 0.958, accuracy of 0.984 and Hausdorff distance of 2.76. The proposed framework is compared with FCN, UNet, Attention UNet, ResUNet, UNet++, Vision Transformer and TransUNet, and the results show its superiority. The results show that the combination of ResUNet++ and Vision Transformers greatly enhances the performance of segmentation, boundary delineation, and generalization in the field of advanced medical image analysis applications.
G. Satyanarayana, Kadali Satyanarayana, Renuka Kondabala et al.· 2026 International Conferenc...· 0 citations
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
Objective Liver cancer is one of the most fatal types of cancer in the world, and early identification is one of the most crucial considerations that may improve the likelihood of recovery. Hence, building a computer-aided diagnosis (CAD) system that improves radiologists’ operations by providing automated, high-precision small-tumor segmentation in complicated medical imaging is the main objective of this research study. Methods Conventional deep learning techniques, like U-Net, have shown impressive overall tumor delineation results, but they frequently struggle to distinguish between large and small tumors. The occurrence of tiny tumors and tumors with different shapes in the intricate liver exacerbates this deficiency. This paper proposed an MSA-Net (Multi-Scale Attention Network), a variant of the U-Net architecture, to address these problems. This is accomplished by the effective extraction of small-tumor features, which is made possible by the architecture’s integration of multi-scale convolutional layers into the encoder and decoder pathways, as well as an attention mechanism that concentrates on salient regions and gathers contextual data across several receptive fields. Results The proposed MSA-Net was trained and tested using the publicly accessible 3DIRCADb and LiTS datasets. The 3Dircadb dataset yielded a dice score of 92.00% and a Jaccard index of 86.00% for small tumors, whereas the LiTS dataset yielded a dice score of 72.57% and a Jaccard index of 65.35% for small tumors. Conclusion In contrast to other research, our technique assesses the outcomes by evaluating large and small tumors independently. The findings demonstrate that MSA-Net greatly improves segmentation, particularly in tiny tumors, demonstrating its usefulness in real-world clinical contexts.
Beenish Hina, Muazzam Maqsood, Asma Sattar et al.· Science in progress· 0 citations
Brain neoplasm is life-threatening abnormalities that require accurate segmentation for effective diagnosis along with treatment planning. Existing segmentation methods often depend on manual annotations or supervised learning, which are time-consuming, subjective, and not scalable. To overcome these challenges, this study proposes a fully unsupervised brain tumor segmentation framework using multimodal MRI data. The 3D MRI volumes are first sliced into 2D axial views and preprocessed to enhance quality. K-means clustering is then applied to generate pseudo-labels that distinguish tumor from healthy regions. These labels guide M-Net CNN training process, which produces coarse tumor localization. A U-Net model further refines the segmentation by fusing the original sliced MRI inputs with M-Net CNN outputs to enhance boundary precision and anatomical alignment. The proposed pipeline achieved a high Dice coefficient of 84.2%, demonstrating its accuracy and robustness. This multi-stage, annotation-free approach enables precise tumor delineation and is suitable for large-scale clinical integration.
J. Adlin, Arockia Selva Saroja· ITEGAM- Journal of Engineeri...· 0 citations