Skin Cancer Diagnosis Using VGG16 and Transfer Learning: Analyzing the Effects of Data Quality over Quantity on Model Efficiency DOI Creative Commons

Khamsa Djaroudib,

Pascal Lorenz,

Rime Belkacem Bouzida

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(17), P. 7447 - 7447

Published: Aug. 23, 2024

The recent increase in the prevalence of skin cancer, along with its significant impact on individuals’ lives, has garnered attention many researchers field deep learning models, especially following promising results observed using these models medical field. This study aimed to develop a system that can accurately diagnose one three types cancer: basal cell carcinoma (BCC), melanoma (MEL), and nevi (NV). Additionally, it emphasizes importance image quality, as studies focus quantity images used learning. In this study, transfer was employed pre-trained VGG-16 model alongside dataset sourced from Kaggle. Three were trained while maintaining same hyperparameters script ensure fair comparison. However, data train each varied observe specific effects hypothesize about quality within highest validation score selected for further testing separate test dataset, which had not seen before, evaluate model’s performance accurately. work contributes existing body research by demonstrating critical role enhancing diagnostic accuracy, providing comprehensive evaluation cancer detection offering insights guide future improvements

Language: Английский

Advancements in deep learning for early diagnosis of Alzheimer’s disease using multimodal neuroimaging: challenges and future directions DOI Creative Commons
Muhammad Liaquat Raza,

Syed Belal Hassan,

Subia Jamil

et al.

Frontiers in Neuroinformatics, Journal Year: 2025, Volume and Issue: 19

Published: May 2, 2025

Introduction Alzheimer’s disease is a progressive neurodegenerative disorder challenging early diagnosis and treatment. Recent advancements in deep learning algorithms applied to multimodal brain imaging offer promising solutions for improving diagnostic accuracy predicting progression. Method This narrative review synthesizes current literature on applications using neuroimaging. The process involved comprehensive search of relevant databases (PubMed, Embase, Google Scholar ClinicalTrials.gov ), selection pertinent studies, critical analysis findings. We employed best-evidence approach, prioritizing high-quality studies identifying consistent patterns across the literature. Results Deep architectures, including convolutional neural networks, recurrent transformer-based models, have shown remarkable potential analyzing neuroimaging data. These models can effectively structural functional modalities, extracting features associated with pathology. Integration multiple modalities has demonstrated improved compared single-modality approaches. also promise predictive modeling, biomarkers forecasting Discussion While approaches show great potential, several challenges remain. Data heterogeneity, small sample sizes, limited generalizability diverse populations are significant hurdles. clinical translation these requires careful consideration interpretability, transparency, ethical implications. future AI neurodiagnostics looks promising, personalized treatment strategies.

Language: Английский

Citations

0
AI with agency: a vision for adaptive, efficient, and ethical healthcare DOI Creative Commons
Víctor Fuentes, Hezerul Abdul Karim, Myles Joshua Toledo Tan

et al.

Frontiers in Digital Health, Journal Year: 2025, Volume and Issue: 7

Published: May 7, 2025

Language: Английский

Citations

0
Optimizing the power of AI for fracture detection: from blind spots to breakthroughs DOI
Shima Behzad, Liesl S. Eibschutz, Max Yang Lu

et al.

Skeletal Radiology, Journal Year: 2025, Volume and Issue: unknown

Published: May 23, 2025

Language: Английский

Citations

0
Deep learning-driven modality imputation and subregion segmentation to enhance high-grade glioma grading DOI Creative Commons

Jiabin Yu,

Qi Liu, Chenjie Xu

et al.

BMC Medical Informatics and Decision Making, Journal Year: 2025, Volume and Issue: 25(1)

Published: May 30, 2025

Language: Английский

Citations

0
RegMamba: An Improved Mamba for Medical Image Registration DOI Open Access

Xin Hu,

Jiaqi Chen, Yilin Chen

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(16), P. 3305 - 3305

Published: Aug. 20, 2024

Deformable medical image registration aims to minimize the differences between fixed and moving images provide comprehensive physiological or structural information for further analysis. Traditional learning-based convolutional network approaches usually suffer from problem of perceptual limitations, in recent years, Transformer architecture has gained popularity its superior long-range relational modeling capabilities, but still faces severe computational challenges handling high-resolution images. Recently, selective state-space models have shown great potential vision domain due their fast inference efficient modeling. Inspired by this, this paper, we propose RegMamba, a novel that combines (SSMs), designed efficiently capture complex correspondence while maintaining effort. Firstly our model introduces Mamba remotely process dependencies data large deformations. At same time, use scaled layer alleviate spatial loss 3D flattening processing Mamba. Then, deformable residual module (DCRM) is proposed adaptively adjust sampling position deformations more flexible features learning fine-grained different anatomical structures construct local correspondences improve perception. We demonstrate advanced performance method on LPBA40 IXI public datasets.

Language: Английский

Citations

2
Improved Segmentation of Cellular Nuclei Using UNET Architectures for Enhanced Pathology Imaging DOI Open Access
S. Alonso Castro, Vítor Pereira, Rui Silva

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(16), P. 3335 - 3335

Published: Aug. 22, 2024

Medical imaging is essential for pathology diagnosis and treatment, enhancing decision making reducing costs, but despite various computational methodologies proposed to improve modalities, further optimization needed broader acceptance. This study explores deep learning (DL) classifying segmenting pathological data, optimizing models accurately predict generalize from training new data. Different CNN U-Net architectures are implemented segmentation tasks, with their performance evaluated on histological image datasets using enhanced pre-processing techniques such as resizing, normalization, data augmentation. These trained, parameterized, optimized metrics accuracy, the DICE coefficient, intersection over union (IoU). The experimental results show that method improves efficiency of cell compared networks, U-NET W-UNET. has improved IoU 0.9077 0.9675, about 7% better results; also, values coefficient obtained 0.9215 0.9916, results, surpassing reported in literature.

Language: Английский

Citations

2
A Short Survey on Multimodal Data Fusion in Image Classification DOI
Toufik Datsi, Khalid Aznag,

Brahim Ait Ben Ali

et al.

Published: April 24, 2024

Advancements in multimodal learning have experienced rapid growth over the past decade, particularly within various domains, with a significant emphasis on developments computer vision. Multimodal data fusion has become increasingly prominent realm of image classification, where integration diverse sources enhances overall understanding and performance classification models. This survey delves into recent strides made field classification. Additionally, paper undertakes comparative study, critically evaluating effectiveness different approaches. The aim is to provide comprehensive overview current state-of-the-art for identify key trends, challenges, opportunities this evolving field.

Language: Английский

Citations

1
Advancements in Computer-Aided Diagnosis of Celiac Disease: A Systematic Review DOI Creative Commons
Ivana Hartmann Tolić, Marija Habijan, Irena Galić

et al.

Biomimetics, Journal Year: 2024, Volume and Issue: 9(8), P. 493 - 493

Published: Aug. 14, 2024

Celiac disease, a chronic autoimmune condition, manifests in those genetically prone to it through damage the small intestine upon gluten consumption. This condition is estimated affect approximately one every hundred individuals worldwide, though often goes undiagnosed. The early and accurate diagnosis of celiac disease (CD) critical preventing severe health complications, with computer-aided diagnostic approaches showing significant promise. However, there shortage review literature that encapsulates field’s current state offers perspective on future advancements. Therefore, this critically assesses role imaging techniques, biomarker analysis, computer models improving CD diagnosis. We highlight strengths advanced non-invasive appeal analyses, while also addressing ongoing challenges standardization integration into clinical practice. Our analysis stresses importance diagnostics fast-tracking CD, highlighting necessity for research refine these effective implementation settings. Future field will focus standardizing CAD protocols broader use exploring genetic protein data enhance detection personalize treatment strategies. These advancements promise improvements patient outcomes implications managing diseases.

Language: Английский

Citations

1
Framework for Deep Learning Based Multi-Modality Image Registration of Snapshot and Pathology Images DOI Creative Commons
R. Schoop, Lotte M. de Roode, Lisanne L. de Boer

et al.

IEEE Journal of Biomedical and Health Informatics, Journal Year: 2024, Volume and Issue: 28(11), P. 6699 - 6711

Published: Aug. 16, 2024

Multi-modality image registration is an important task in medical imaging because it allows for information from different domains to be correlated. Histopathology plays a crucial role oncologic surgery as the gold standard investigating tissue composition surgically excised specimens. Research studies are increasingly focused on registering modalities such white light cameras, magnetic resonance imaging, computed tomography, and ultrasound pathology images. The main challenge tasks involving images comes addressing considerable amount of deformation present. This work provides framework deep learning-based multi-modality microscopic another modality. proposed validated prostate ex-vivo camera snapshot hematoxylin-eosin same specimen. A pipeline presented detailing data acquisition, protocol considerations, dissimilarity, training experiments, validation. comprehensive analysis done impact pre-processing, augmentation, loss functions, regularization. supplemented by clinically motivated evaluation metrics avoid pitfalls only using ubiquitous comparison metrics. Consequently, robust configuration capable performing desired found. Utilizing approach, we achieved dice similarity coefficient 0.96, mutual score 0.54, target error 2.4 mm, regional 0.70.

Language: Английский

Citations

1
Traditional and advanced AI methods used in the area of neuro-oncology DOI
Soumyaranjan Panda,

Suman Sourav Biswal,

Sarit Samyak Rath

et al.

Elsevier eBooks, Journal Year: 2024, Volume and Issue: unknown, P. 277 - 300

Published: Oct. 25, 2024

Language: Английский

Citations

1