International Journal of Imaging Systems and Technology最新文献

{"title":"Modern Academia: From “Publish or Perish” to “Monetize or Collapse”","authors":"Mohamed L. Seghier, Mahmoud Meribout","doi":"10.1002/ima.70137","DOIUrl":"https://doi.org/10.1002/ima.70137","url":null,"abstract":"<p>Academia needs money and it needs a lot and soon! Recent reports from many countries revealed that modern academia is grappling with a significant crisis in sustaining its core mission financially without burdening students with high tuition fees or relying heavily on governmental funders or private donors. This trend is more pronounced in countries like the UK than in the USA, with its strong university-industry partnerships (e.g., Silicon Valley), or in China and many European countries where universities are supported by their governments. However, with substantial cuts to government budgets for higher education, public funding is rapidly depleting, necessitating the urgent development of alternative funding models. For instance, the recent threats to some UK universities at the risk of closing whole departments and the huge loss in funding in the US to some universities and agencies [<span>1</span>] should serve as a wake-up call for all stakeholders to prevent academia from going broke. In these uncertain times, universities are asked to make further drastic cuts or merge just to survive [<span>2</span>].</p><p>The UK provides one example to gauge the true impact of financial turmoil on academia. For example, more than half of income of UK universities come from tuition fees, predominantly from international students, while a seventh of the income come from research grants (government bodies or charities) [<span>2</span>]. According to the UK’ Office for Students, and despite an income of tens of billions of dollars, 40% of England's universities are expected to run budget deficits this year, with more than 70 universities in the UK have announced staff redundancies, department closures, programs phasing out, and other forms of restructuring [<span>3</span>]. A similar alarming picture is also emerging in the US with research programs been closed in particular in domains judged not important by the new policy makers, as well as universities targeted with drastic cuts for not aligning with government's positions and policies [<span>4</span>].</p><p>Three traditional models are gaining momentum in the current climate to save academia: expanding partnerships with industry, promoting a new breed of academic entrepreneurs, and monetising academic expertise. These models, though not new, are being administered to academia at high pace and with a degree of urgency. For instance, over the past decades, the emphasis on industrial collaborations in grant applications for science and engineering disciplines has expanded significantly, growing from a few statements to full pages. As a result, new terminology such as market maps, technological readiness levels, cost savings, competitiveness, spin-offs, patents, and marketization has become commonplace in these grant applications. Likewise, universities are establishing more incubators and frameworks to encourage their academic staff to transform their ideas and innovative solutions into marketable pro","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 4","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ima.70137","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144197402","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hybrid Support Vector Machine-Convolutional Neural Networks Multi-Classification Models for Detection of Kidney Stones","authors":"Setlhabi Letlhogonolo Rapelang,&nbsp;Ibidun Christiana Obagbuwa","doi":"10.1002/ima.70128","DOIUrl":"https://doi.org/10.1002/ima.70128","url":null,"abstract":"<p>The accurate and early detection of kidney stones is crucial for effective treatment and patient management. This study presents a hybrid machine learning approach combining Support Vector Machines (SVM) and Convolutional Neural Networks (CNN) for the multi-classification of kidney stones. The proposed model leverages the feature extraction capabilities of CNNs with the robust classification performance of SVMs to improve diagnostic accuracy. The methodology is validated on a publicly available kidney stone dataset, and the experimental results demonstrate the superiority of the hybrid model over standalone CNN and SVM models. Different techniques, such as enhancing the contrast of the images, gray conversion to train with one channel, Gaussian filter to blur the noise of the images, data augmentation, and SMOTE to balance the dataset, using 5-fold cross-validation to prevent overfitting. Features that we extracted from CNN were optimized and classified using SVM, KNN, and RF. All the classifiers we incorporated showed a high overall classification accuracy of over 98%. Among these classifiers, the proposed Hybrid CNN-SVM model outperformed other models with a higher overall test accuracy of 98.49%. At the same time, CNN-KNN, CNN-RF, and CNN achieved an accuracy of 98.46%, 98.01%, and 97.62%, respectively. These classifiers show the effectiveness of hybrid models in reducing training time and improving classification accuracy compared to single models.</p>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 4","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ima.70128","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144190670","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A 3D-2D Rigid Liver Registration Method Using Pre-Training and Transfer Learning With Staged Alignment of Anatomical Landmarks","authors":"Junchen Hao,&nbsp;Baochun He,&nbsp;Yue Dai,&nbsp;Yuchong Li,&nbsp;Yu Wang,&nbsp;Rui Zhao,&nbsp;Ruoqi Lian,&nbsp;Xiaojun Zeng,&nbsp;Haisu Tao,&nbsp;Jian Yang,&nbsp;Chihua Fang,&nbsp;Huiyan Jiang,&nbsp;Fucang Jia","doi":"10.1002/ima.70124","DOIUrl":"https://doi.org/10.1002/ima.70124","url":null,"abstract":"<div>\u0000 \u0000 <p>Augmented reality navigation in laparoscopic liver resection can integrate surgical planning information such as liver resection lines, blood vessels, and tumors to enhance surgical safety. However, the 3D-2D registration still faces challenges, including long registration time and manual initialization. Preoperative 3D liver point cloud and intraoperative laparoscopic image data are pre-trained to generate a patient-specific initial pose. A staged fine registration strategy targeting local anatomical landmarks is employed, involving normalization of the distance loss between the projection points of various anatomical landmarks in the preoperative 3D model and the corresponding ground truth landmarks in the intraoperative 2D laparoscopic images. The proposed method was evaluated using pixel-wise reprojection error (RPE) and target registration error (TRE). The results demonstrate that the method achieves superior registration accuracy compared to existing rigid registration methods. Deep learning integrated into 3D-2D rigid registration achieved full automation and sped up the computation.</p>\u0000 </div>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 4","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144190671","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GSCCANet: Dual Decoder Network Fusing Edge Focus and Global Channel Attention for Precise Segmentation of Colonic Polyps","authors":"Jun Su,&nbsp;Xinyi Chen,&nbsp;Orest Kochan,&nbsp;Mariana Levkiv,&nbsp;Krzysztof Przystupa","doi":"10.1002/ima.70129","DOIUrl":"https://doi.org/10.1002/ima.70129","url":null,"abstract":"<div>\u0000 \u0000 <p>Colorectal cancer is the second most common cancer globally. Its high mortality necessitates early polyp detection to mitigate the risk of the disease. However, conventional segmentation methods are susceptible to noise interference and have a limited accuracy in complex environments. To address these challenges, we propose GSCCANet with an encoder-dual decoder co-design. The encoder employs hybrid Transformer (MiT) for efficient multi-scale global feature extraction. Dual decoders collaborate via SAFM and REF-RA modules to enhance segmentation precision through global semantics and boundary refinement. In particular, SAFM enhances lesion coherence via channel-space attention fusion, while REF-RA strengthens low-contrast edge response using high-frequency gradients and reverse attention, optimized through progressive fusion. Additionally, combined Focal Loss and Weighted IoU Loss mitigate the problem of undetected small polyps. Experiments on five datasets show GSCCANet surpasses baselines. It achieves 94.7% mDice and 90.1% mIoU on CVC-ClinicDB (regular) and 80.1% mDice and 72.5% mIoU on ETIS-LaribPolypDB (challenging). Cross-domain tests (CVC-ClinicDB <span></span><math>\u0000 <semantics>\u0000 <mrow>\u0000 <mo>→</mo>\u0000 </mrow>\u0000 <annotation>$$ to $$</annotation>\u0000 </semantics></math> Kvasir) confirm strong adaptability with 0.2% mDice fluctuation. These results prove that GSCCANet offers high-precision and generalizable solutions through global–local synergy, edge enhancement, and efficient computation.</p>\u0000 </div>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 4","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144171238","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Performance Indicator for Optimizing Source–Detector Separation in Functional Near-Infrared Spectroscopy","authors":"Serhat Ilgaz Yoner,&nbsp;Gokhan Ertas","doi":"10.1002/ima.70113","DOIUrl":"https://doi.org/10.1002/ima.70113","url":null,"abstract":"<p>The performance of Functional Near-Infrared Spectroscopy (fNIRS) devices critically depends on the probe design, which affects signal quality, spatial and depth resolution, and data reliability. A critical component of probe separation is source-to-detector separation, which is defined as the distance between the light source and the detector. Optimizing this separation is essential for improving the signal-to-noise ratio (SNR) and sensitivity at depth (SAD). Larger separations enhance depth resolution, facilitating more accurate assessments of brain activity. Conversely, excessive separation may reduce SNR due to the lower light intensity received by the detector. In this study, a performance indicator was created to optimize separation by integrating the SNR and SAD. A probe was constructed that featured one light source and four detectors mounted on a mechanism that allowed for adjustable separations. A phantom mimicking brain tissue was used. Signals were recorded from the probe positioned on the phantom at various separations, employing light sources emitting light at wavelengths of 730, 800, and 850 nm, and optical power levels of 19, 26, 32, 38, and 44 mW. The SNR values for each separation were computed from the recorded signals, whereas the SAD values were obtained from existing literature. The performance indicator was developed as a weighted sum of SNR and SAD, normalized between 0 and 1, with higher values indicating enhanced probe performance due to optimized separation. The indicator is expected to improve the reliability of fNIRS data; however, further research involving diverse populations is required to validate its practical application.</p>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 4","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ima.70113","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144171612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing Melanoma Detection With Anisotropic Median Filtering and Multinomial Classification Vision Transformer","authors":"R. Naga Swetha,&nbsp;Vimal K. Shrivastava","doi":"10.1002/ima.70119","DOIUrl":"https://doi.org/10.1002/ima.70119","url":null,"abstract":"<div>\u0000 \u0000 <p>Skin cancer is one of the most prevalent and dangerous types of cancer globally, caused by unrepaired DNA damage leading to abnormal cell growth in the epidermis. Melanoma, in particular, is one of the most hazardous forms, requiring early and precise diagnosis to improve patient outcomes. Early detection and diagnosis are vital for reducing the mortality rates associated with this aggressive cancer. In this paper, we propose a novel approach that combines an anisotropic median filter (AMF) with a modified vision transformer, termed the Multinomial Classification Vision Transformer (MCVT) for skin cancer classification. The AMF is used as pre-processing to effectively remove noise and enhance image quality, preserving critical features essential for accurate classification. On the other hand, the MCVT leverages its robust feature extraction capabilities to classify melanoma with high accuracy. We utilized the HAM10000 dataset for training and evaluation. Our proposed method outperforms existing state-of-the-art techniques, achieving an overall classification accuracy of 91% and a melanoma classification accuracy of 89%. These results demonstrate the potential of integrating AMF and MCVT to enhance skin cancer classification, with a particular focus on improving melanoma detection.</p>\u0000 </div>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 3","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144135581","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimize Neural Fuzzy Systems for High-Dimensional Breast Cancer Data Analysis: A Deep Learning Approach","authors":"Jingjing Jin,&nbsp;Yunhu Huang","doi":"10.1002/ima.70117","DOIUrl":"https://doi.org/10.1002/ima.70117","url":null,"abstract":"<div>\u0000 \u0000 <p>Accurate and timely analysis of breast cancer data is crucial for the successful deployment and advancement of intelligent healthcare systems. Traditional health status prediction methods, which often rely on shallow models, fall short in complex clinical scenarios and are still unsatisfying for many real-world applications. This situation has inspired us to propose a deep learning-enhanced framework for health data flow prediction. The paper introduces a new three-layer soft computing method for predicting health status using optimizing neural fuzzy systems (ONFS). This approach enhances interpretability by considering spatial correlations in medical data. We start with feature selection based on the Pearson correlation coefficient (PCC) to eliminate variables with minimal linear or nonlinear relationships. Next, subtractive clustering optimization is applied in each layer to refine the system parameters simultaneously. The ONFS offers clearer and more straightforward explanations of health features in high-dimensional data analysis. Experimental results demonstrate the superiority of ONFS over existing methods, achieving an average RMSE reduction of 17.2% and a 98% reduction in rules compared to SVM, with competitive computational efficiency. This research underscores the potential of deep learning-augmented ONFS in enhancing breast cancer data analysis, supporting the information science objectives of precision and interpretability in healthcare data processing.</p>\u0000 </div>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 3","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144135578","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-Organ Segmentation Network for Female Pelvic MR Images Based on Hierarchical Decoupled Fusion and Multi-Scale Feature Processing","authors":"Xiaobao Liu,&nbsp;Cheng Zhang,&nbsp;Wenjuan Gu,&nbsp;Tingqiang Yao,&nbsp;Jihong Shen,&nbsp;Dan Tang","doi":"10.1002/ima.70122","DOIUrl":"https://doi.org/10.1002/ima.70122","url":null,"abstract":"<div>\u0000 \u0000 <p>Accurate segmentation of the female pelvic structures is crucial for diagnosing and treating female pelvic disorders. However, the complex and variable shapes of the pelvic organs, along with their blurred boundary features, present significant challenges. To address the aforementioned issues, HDFMFP-Net, a multi-organ segmentation network for female pelvic MR images, is constructed based on hierarchical decoupled fusion and multi-scale feature processing. Firstly, to address the problem of the complex and variable shapes of the pelvic organs, a hierarchical decoupled fusion module is constructed as a fundamental unit. This module is used to build a lightweight encoder-decoder structure, which reduces the accumulation of estimated bias while effectively extracting the complex boundary features of pelvic organs through multi-scale feature learning and cross-layer information interaction. Secondly, to address the challenge posed by the blurred boundary features of pelvic organs, a multi-scale feature processing module is integrated into the skip connections. This module captures features at multiple scales from both local and global perspectives, enhancing the network's ability to represent the fuzzy boundary features of organs. Finally, the algorithm was evaluated on the female pelvic dataset to segment the bladder, uterus, and rectum in female pelvic images, and the mIoU, mDice, and mPA of HDFMFP-Net reached 92.98%, 96.3%, and 96.24%, respectively, with a model size of only 2.80 M. The results demonstrate that the proposed method offers a promising approach for the automatic segmentation of female pelvic organs.</p>\u0000 </div>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 3","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144135580","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhanced Segmentation of Active and Nonactive Multiple Sclerosis Plaques in T1 and FLAIR MRI Images Using Transformer-Based Encoders","authors":"Mahsa Naeeni Davarani,&nbsp;Ali Arian Darestani,&nbsp;Virginia Guillen Cañas,&nbsp;Mohammad Hossein Harirchian,&nbsp;Amin Zarei,&nbsp;Sanaz Heydari Havadaragh,&nbsp;Hasan Hashemi","doi":"10.1002/ima.70120","DOIUrl":"https://doi.org/10.1002/ima.70120","url":null,"abstract":"<p>Demyelinating plaques in multiple sclerosis (MS) can be visualized using magnetic resonance imaging (MRI), where accurate segmentation of active and nonactive lesions is critical for diagnosis, monitoring disease progression, and guiding treatment. Fluid-attenuated inversion recovery )FLAIR( images are widely used to detect both types of lesions, while T1-weighted images are, particularly, useful for identifying active plaques, although they are more challenging to segment due to their lower contrast and smaller lesion size. To enhance the segmentation accuracy of MS plaques, focusing on both active and non-active lesions, by utilizing TransUNet, a transformer-based neural network. The model's performance is evaluated on T1-weighted and FLAIR MRI images, with a specific focus on improving the segmentation of active plaques in T1-weighted images, which are traditionally more difficult to segment. The dataset included MRI scans from 174 patients diagnosed with MS, a significant expansion compared to previous studies. Additionally, 21 external subject test data were used to validate the model's generalizability. TransUNet was applied separately to T1-weighted and FLAIR images. Preprocessing steps included skull stripping and normalization. The model's performance was assessed using standard evaluation metrics, including Dice Coefficient, sensitivity, specificity, intersection over union (IoU), and Hausdorff distance at 95% (HD95). The study also conducted a comparative analysis between TransUNet and the widely used nnU-Net model. For FLAIR images, TransUNet achieved a sensitivity of 0.763, specificity of 0.998, IoU of 0.563, Dice coefficient of 0.712, and HD95 of 5.402 mm on the internal test set. On the external test set, it maintained a sensitivity of 0.739, specificity of 0.999, IoU of 0.551, Dice coefficient of 0.704, and HD95 of 14.630 mm. For T1-weighted images, the model showed a sensitivity of 0.494, specificity of 1.000, IoU of 0.411, Dice coefficient of 0.548, and HD95 of 22.144 mm on the internal test set. On the external test set, it improved to a sensitivity of 0.725, specificity of 0.999, IoU of 0.573, Dice coefficient of 0.693, and HD95 of 5.146 mm. Compared to nnU-Net on FLAIR images, TransUNet achieved a higher Dice coefficient (0.712 vs. 0.710) and significantly lower HD95 (5.402 vs. 28.300 mm). TransUNet significantly outperforms traditional methods, particularly in FLAIR images, demonstrating improved accuracy and boundary delineation. While T1-weighted images present challenges, the model shows potential for refinement. This study highlights the effectiveness of transformer-based architectures in medical image segmentation, suggesting TransUNet as a valuable tool for MS diagnosis and treatment monitoring.</p>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 3","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ima.70120","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144135579","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DSSViT: Multi-Scale Adaptive Fusion Vision Transformer With Dense Feature Reuse for Robust Pneumonia Detection in Chest Radiography","authors":"Jinhui Huang","doi":"10.1002/ima.70127","DOIUrl":"https://doi.org/10.1002/ima.70127","url":null,"abstract":"<div>\u0000 \u0000 <p>Accurate pneumonia diagnosis using chest x-rays (CXR) remains a critical challenge due to the need for precise extraction of fine-grained local features and effective multi-scale spatial pattern recognition. While Vision Transformer (ViT) models have demonstrated strong performance in medical imaging, they often struggle with these aspects, limiting their effectiveness in clinical applications. This study proposes Dense-SEA ViT (DSSViT), an enhanced Vision Transformer architecture, to address these limitations by improving fine-grained feature representation and multi-scale spatial information capture for pneumonia detection. DSSViT integrates DenseNet121 as a feature extractor to enhance feature reuse and improve information flow, thereby compensating for ViT's weakness in capturing low-level visual details. Additionally, the Squeeze-Excitation and Adaptive Fusion (SEA) mechanism is introduced to calibrate channel attention and enable multi-scale adaptive fusion, enhancing the model's ability to extract critical diagnostic features while reducing noise interference. The proposed architecture was evaluated on a chest X-ray dataset for pneumonia classification. Experimental results demonstrate that DSSViT achieves superior feature extraction capability, leading to a test accuracy of 97.69%, outperforming baseline models such as EfficientNet (93.90%) and VGG19 (96.57%). These findings suggest that DSSViT is a promising approach for improving automated pneumonia diagnosis in clinical settings.</p>\u0000 </div>","PeriodicalId":14027,"journal":{"name":"International Journal of Imaging Systems and Technology","volume":"35 3","pages":""},"PeriodicalIF":3.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144140478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}