A precise model for skin cancer diagnosis using hybrid U-Net and improved MobileNet-V3 with hyperparameters optimization

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Feb. 21, 2024

Abstract Skin cancer is a frequently occurring and possibly deadly disease that necessitates prompt precise diagnosis in order to ensure efficacious treatment. This paper introduces an innovative approach for accurately identifying skin by utilizing Convolution Neural Network architecture optimizing hyperparameters. The proposed aims increase the precision efficacy of recognition consequently enhance patients' experiences. investigation tackle various significant challenges recognition, encompassing feature extraction, model design, utilizes advanced deep-learning methodologies extract complex features patterns from images. We learning procedure deep integrating Standard U-Net Improved MobileNet-V3 with optimization techniques, allowing differentiate malignant benign cancers. Also substituted crossed-entropy loss function Mobilenet-v3 mathematical framework bias accuracy. model's squeeze excitation component was replaced practical channel attention achieve parameter reduction. Integrating cross-layer connections among Mobile modules has been leverage synthetic effectively. dilated convolutions were incorporated into receptive field. hyperparameters utmost importance improving efficiency models. To fine-tune hyperparameter, we employ sophisticated methods such as Bayesian method using pre-trained CNN MobileNet-V3. compared existing models, i.e., MobileNet, VGG-16, MobileNet-V2, Resnet-152v2 VGG-19 on “HAM-10000 Melanoma Cancer dataset". empirical findings illustrate optimized hybrid outperforms detection segmentation techniques based high 97.84%, sensitivity 96.35%, accuracy 98.86% specificity 97.32%. enhanced performance this research resulted timelier more diagnoses, potentially contributing life-saving outcomes mitigating healthcare expenditures.

Language: Английский

---

Deep Learning and Optimization-Based Methods for Skin Lesions Segmentation: A Review

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 85467 - 85488

Published: Jan. 1, 2023

Skin cancer is a senior public health issue that could profit from computer-aided diagnosis to decrease the encumbrance of this widespread disease. Researchers have been more motivated develop systems because visual examination wastes time. The initial stage in skin lesion analysis segmentation, which might assist following categorization task. It difficult task sometimes whole be same colors, and borders pigment regions can foggy. Several studies effectively handled segmentation; nevertheless, developing new methodologies improve efficiency necessary. This work thoroughly analyzes most advanced algorithms methods for segmentation. review begins with traditional segmentation techniques, followed by brief using deep learning optimization techniques. main objective highlight strengths weaknesses wide range algorithms. Additionally, it examines various commonly used datasets lesions metrics evaluate performance these

Language: Английский

---

Medical Image Classifications Using Convolutional Neural Networks: A Survey of Current Methods and Statistical Modeling of the Literature

Machine Learning and Knowledge Extraction, Journal Year: 2024, Volume and Issue: 6(1), P. 699 - 736

Published: March 21, 2024

In this review, we compiled convolutional neural network (CNN) methods which have the potential to automate manual, costly and error-prone processing of medical images. We attempted provide a thorough survey improved architectures, popular frameworks, activation functions, ensemble techniques, hyperparameter optimizations, performance metrics, relevant datasets data preprocessing strategies that can be used design robust CNN models. also machine learning algorithms for statistical modeling current literature uncover latent topics, method gaps, prevalent themes future advancements. The results indicate temporal shift in favor designs, such as from use architecture CNN-transformer hybrid. insights point surge practitioners into imaging field, partly driven by COVID-19 challenge, catalyzed detecting diagnosing pathological conditions. This phenomenon likely contributed sharp increase number publications on CNNs imaging, both during after pandemic. Overall, existing has certain gaps scope with respect optimization architectures specifically imaging. Additionally, there is lack post hoc explainability models slow progress adopting low-resource review ends list open research questions been identified through recommendations potentially help set up more robust, reproducible experiments

Language: Английский

---

Fusion of transformer attention and CNN features for skin cancer detection

Applied Soft Computing, Journal Year: 2024, Volume and Issue: 164, P. 112013 - 112013

Published: July 18, 2024

Language: Английский

---

Enhanced Skin Lesion Classification Using Deep Learning, Integrating with Sequential Data Analysis: A Multiclass Approach

Published: Jan. 7, 2025

In dermatological research, accurately identifying different types of skin lesions, such as nodules, is essential for early diagnosis and effective treatment. This study introduces a novel method classifying including by combining unified attention (UA) network with deep convolutional neural networks (DCNNs) feature extraction. The UA processes sequential data, patient histories, while long short-term memory (LSTM) track nodule progression. Additionally, Markov random fields (MRFs) enhance pattern recognition. integrated system classifies lesions evaluates whether they are responding to treatment or worsening, achieving 93% accuracy in distinguishing melanoma, basal cell carcinoma. outperforms existing methods precision sensitivity, offering advancements diagnostics.

Language: Английский

---

Medical image identification methods: A review

Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 169, P. 107777 - 107777

Published: Dec. 5, 2023

Language: Английский

---

FPN-SE-ResNet Model for Accurate Diagnosis of Kidney Tumors Using CT Images

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(17), P. 9802 - 9802

Published: Aug. 30, 2023

Kidney tumors are a significant health concern. Early detection and accurate segmentation of kidney crucial for timely effective treatment, which can improve patient outcomes. Deep learning techniques, particularly Convolutional Neural Networks (CNNs), have shown great promise in medical image analysis, including identifying segmenting tumors. Computed tomography (CT) scans kidneys aid tumor assessment morphology studies, employing semantic techniques precise pixel-level identification surrounding anatomical structures. This paper proposes Squeeze-and-Excitation-ResNet (SE-ResNet) model by combining the encoder stage SE-ResNet with Feature Pyramid Network (FPN). The performance proposed is evaluated using Intersection over Union (IoU) F1-score metrics. Experimental results demonstrate that models achieve impressive IoU scores background, kidney, segmentation, mean ranging from 0.988 to 0.981 Seresnet50 Seresnet18, respectively. Notably, exhibits highest score segmentation. These findings suggest accurately identify segment regions interest CT images renal carcinoma, higher versions generally exhibiting superior performance. good tool classification, aiding professionals early diagnosis intervention.

Language: Английский

---

Emotional State Detection Using Electroencephalogram Signals: A Genetic Algorithm Approach

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(11), P. 6394 - 6394

Published: May 23, 2023

Emotion recognition based on electroencephalogram signals (EEG) has been analyzed extensively in different applications, most of them using medical-grade equipment laboratories. The trend human-centered artificial intelligence applications is toward portable sensors with reduced size and improved portability that can be taken to real life scenarios, which requires systems efficiently analyze information time. Currently, there no specific set features or number electrodes defined classify emotions EEG signals, performance may the combination all available but could result high dimensionality even worse performance; solve problem dimensionality, this paper proposes use genetic algorithms (GA) automatically search optimal subset data for emotion classification. Publicly 2548 describing waves related emotional states are analyzed, then 49 algorithms. results show only out sufficient create machine learning (ML) classification models with, such as k-nearest neighbor (KNN), random forests (RF) neural networks (ANN), obtaining 90.06%, 93.62% 95.87% accuracy, respectively, higher than 87.16% 89.38% accuracy previous works.

Language: Английский

---

Melanoma identification and classification model based on fine-tuned convolutional neural network

Digital Health, Journal Year: 2024, Volume and Issue: 10

Published: Jan. 1, 2024

Breakthroughs in skin cancer diagnostics have resulted from recent image recognition and Artificial Intelligence (AI) technology advancements. There has been growing that can be lethal to humans. For instance, melanoma is the most unpredictable terrible form of cancer.

Language: Английский

---

Research on Tool Remaining Life Prediction Method Based on CNN-LSTM-PSO

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 80448 - 80464

Published: Jan. 1, 2023

Efficient and accurate prediction of tool Remaining Useful Life (RUL) is the key to improve product accuracy, work efficiency reduce machining costs. Aiming at problems weak wear state features, difficult extraction, low precision this study proposes a CNN-LSTM-PSO remaining life method based on multi-channel feature fusion.Firstly, computer vision, information fusion technology, multi-source sensor signals collected during cycle are effectively processed analyzed, sample data set spatio-temporal correlation traffic flow constructed. Secondly, was input into model, CNN network obtained sequence vector by extracting spatial characteristics data, multi-layer LSTM extract time-dependent PSO algorithm optimized hyperparameters in CNN-LSTM model. The accuracy RUL model fitting further improved. results show that can predict wear, with mean absolute error (MAE) value 1.0892, root square (RMSE) 1.3520, determination coefficient R ² 0.9961; Through comparative analysis ablation experiments, it found proposed has highest lowest values MAE RMSE, closest 1, which certain advantages.The reference engineering practical significance for related research residual prediction.

Language: Английский

---