Deep Learning for Medical Image-Based Cancer Diagnosis

Cancers, Год журнала: 2023, Номер 15(14), С. 3608 - 3608

Опубликована: Июль 13, 2023

(1) Background: The application of deep learning technology to realize cancer diagnosis based on medical images is one the research hotspots in field artificial intelligence and computer vision. Due rapid development methods, requires very high accuracy timeliness as well inherent particularity complexity imaging. A comprehensive review relevant studies necessary help readers better understand current status ideas. (2) Methods: Five radiological images, including X-ray, ultrasound (US), computed tomography (CT), magnetic resonance imaging (MRI), positron emission (PET), histopathological are reviewed this paper. basic architecture classical pretrained models comprehensively reviewed. In particular, advanced neural networks emerging recent years, transfer learning, ensemble (EL), graph network, vision transformer (ViT), introduced. overfitting prevention methods summarized: batch normalization, dropout, weight initialization, data augmentation. image-based analysis sorted out. (3) Results: Deep has achieved great success diagnosis, showing good results image classification, reconstruction, detection, segmentation, registration, synthesis. However, lack high-quality labeled datasets limits role faces challenges rare multi-modal fusion, model explainability, generalization. (4) Conclusions: There a need for more public standard databases cancer. pre-training potential be improved, special attention should paid multimodal fusion supervised paradigm. Technologies such ViT, few-shot will bring surprises images.

Язык: Английский

---

CerCan·Net: Cervical cancer classification model via multi-layer feature ensembles of lightweight CNNs and transfer learning

Expert Systems with Applications, Год журнала: 2023, Номер 229, С. 120624 - 120624

Опубликована: Июнь 2, 2023

Язык: Английский

---

Applied Artificial Intelligence in Healthcare: A Review of Computer Vision Technology Application in Hospital Settings

Journal of Imaging, Год журнала: 2024, Номер 10(4), С. 81 - 81

Опубликована: Март 28, 2024

Computer vision (CV), a type of artificial intelligence (AI) that uses digital videos or sequence images to recognize content, has been used extensively across industries in recent years. However, the healthcare industry, its applications are limited by factors like privacy, safety, and ethical concerns. Despite this, CV potential improve patient monitoring, system efficiencies, while reducing workload. In contrast previous reviews, we focus on end-user CV. First, briefly review categorize other (job enhancement, surveillance automation, augmented reality). We then developments hospital setting, outpatient, community settings. The advances monitoring delirium, pain sedation, deterioration, mechanical ventilation, mobility, surgical applications, quantification workload hospital, for events outside highlighted. To identify opportunities future also completed journey mapping at different levels. Lastly, discuss considerations associated with outline processes algorithm development testing limit expansion healthcare. This comprehensive highlights ideas expanded use

Язык: Английский

---

Skin-CAD: Explainable deep learning classification of skin cancer from dermoscopic images by feature selection of dual high-level CNNs features and transfer learning

Computers in Biology and Medicine, Год журнала: 2024, Номер 178, С. 108798 - 108798

Опубликована: Июнь 25, 2024

Язык: Английский

---

Enhanced image diagnosing approach in medicine using quantum adaptive machine learning techniques

Optical and Quantum Electronics, Год журнала: 2024, Номер 56(4)

Опубликована: Янв. 30, 2024

Язык: Английский

---

Robust Multi-Subtype Identification of Breast Cancer Pathological Images Based on a Dual-Branch Frequency Domain Fusion Network

Sensors, Год журнала: 2025, Номер 25(1), С. 240 - 240

Опубликована: Янв. 3, 2025

Breast cancer (BC) is one of the most lethal cancers worldwide, and its early diagnosis critical for improving patient survival rates. However, extraction key information from complex medical images attainment high-precision classification present a significant challenge. In field signal processing, texture-rich typically exhibit periodic patterns structures, which are manifested as energy concentrations at specific frequencies in frequency domain. Given above considerations, this study designed to explore application domain analysis BC histopathological classification. This proposes dual-branch adaptive fusion network (AFFNet), enable each branch specialize distinct features pathological images. Additionally, two different approaches, namely Multi-Spectral Channel Attention (MSCA) Fourier Filtering Enhancement Operator (FFEO), employed enhance texture minimize loss. Moreover, contributions branches stages dynamically adjusted by frequency-domain-adaptive strategy accommodate complexity multi-scale The experimental results, based on public image datasets, corroborate idea that AFFNet outperforms 10 state-of-the-art methods, underscoring effectiveness superiority

Язык: Английский

---

Lung and Colon Cancer Classification Using Multiscale Deep Features Integration of Compact Convolutional Neural Networks and Feature Selection

Technologies, Год журнала: 2025, Номер 13(2), С. 54 - 54

Опубликована: Фев. 1, 2025

The automated and precise classification of lung colon cancer from histopathological photos continues to pose a significant challenge in medical diagnosis, as current computer-aided diagnosis (CAD) systems are frequently constrained by their dependence on singular deep learning architectures, elevated computational complexity, ineffectiveness utilising multiscale features. To this end, the present research introduces CAD system that integrates several lightweight convolutional neural networks (CNNs) with dual-layer feature extraction selection overcome aforementioned constraints. Initially, it extracts attributes two separate layers (pooling fully connected) three pre-trained CNNs (MobileNet, ResNet-18, EfficientNetB0). Second, uses benefits canonical correlation analysis for dimensionality reduction pooling layer reduce complexity. In addition, features encapsulate both high- low-level representations. Finally, benefit multiple network architectures while reducing proposed merges dual variables then applies variance (ANOVA) Chi-Squared most discriminative integrated CNN architectures. is assessed LC25000 dataset leveraging eight distinct classifiers, encompassing various Support Vector Machine (SVM) variants, Decision Trees, Linear Discriminant Analysis, k-nearest neighbours. experimental results exhibited outstanding performance, attaining 99.8% accuracy cubic SVM classifiers employing merely 50 ANOVA-selected features, exceeding performance individual markedly diminishing framework’s capacity sustain exceptional limited set renders especially advantageous clinical applications where diagnostic precision efficiency critical. These findings confirm efficacy multi-CNN, multi-layer methodology enhancing mitigating constraints systems.

Язык: Английский

---

Multitask Deep Learning-Based Pipeline for Gas Leakage Detection via E-Nose and Thermal Imaging Multimodal Fusion

Chemosensors, Год журнала: 2023, Номер 11(7), С. 364 - 364

Опубликована: Июнь 28, 2023

Innovative engineering solutions that are efficient, quick, and simple to use crucial given the rapid industrialization technology breakthroughs in Industry 5.0. One of areas receiving attention is rise gas leakage accidents at coal mines, chemical companies, home appliances. To prevent harm both environment human lives, automated detection identification type necessary. Most previous studies used a single mode data perform process. However, instead using source/mode, multimodal sensor fusion offers more accurate results. Furthermore, majority individual feature extraction approaches extract either spatial or temporal information. This paper proposes deep learning-based (DL) pipeline combine acquired via infrared (IR) thermal imaging an array seven metal oxide semiconductor (MOX) sensors forming electronic nose (E-nose). The proposed based on three convolutional neural networks (CNNs) models for bidirectional long-short memory (Bi-LSTM) detection. Two used, including intermediate multitask fusion. Discrete wavelet transform (DWT) utilized features extracted from each CNN, providing spectral–temporal representation. In contrast, fusion, discrete cosine (DCT) merge all obtained CNNs trained with data. results show approach has boosted performance reaching accuracy 98.47% 99.25% respectively. These indicate superior Therefore, system capable detecting accurately could be industrial applications.

Язык: Английский

---

MonDiaL-CAD: Monkeypox diagnosis via selected hybrid CNNs unified with feature selection and ensemble learning

Digital Health, Год журнала: 2023, Номер 9

Опубликована: Янв. 1, 2023

Recently, monkeypox virus is slowly evolving and there are fears it will spread as COVID-19. Computer-aided diagnosis (CAD) based on deep learning approaches especially convolutional neural network (CNN) can assist in the rapid determination of reported incidents. The current CADs were mostly an individual CNN. Few employed multiple CNNs but did not investigate which combination has a greater impact performance. Furthermore, they relied only spatial information features to train their models. This study aims construct CAD tool named "Monkey-CAD" that address previous limitations automatically diagnose rapidly accurately.Monkey-CAD extracts from eight then examines best possible influence classification. It employs discrete wavelet transform (DWT) merge diminishes fused features' size provides time-frequency demonstration. These sizes further reduced via entropy-based feature selection approach. finally used deliver better representation input feed three ensemble classifiers.Two freely accessible datasets called Monkeypox skin image (MSID) lesion (MSLD) this study. Monkey-CAD could discriminate among cases with without achieving accuracy 97.1% for MSID 98.7% MSLD respectively.Such promising results demonstrate be health practitioners. They also verify fusing selected boost

Язык: Английский

---

ADHD-AID: Aiding Tool for Detecting Children’s Attention Deficit Hyperactivity Disorder via EEG-Based Multi-Resolution Analysis and Feature Selection

Biomimetics, Год журнала: 2024, Номер 9(3), С. 188 - 188

Опубликована: Март 20, 2024

The severe effects of attention deficit hyperactivity disorder (ADHD) among adolescents can be prevented by timely identification and prompt therapeutic intervention. Traditional diagnostic techniques are complicated time-consuming because they subjective-based assessments. Machine learning (ML) automate this process prevent the limitations manual evaluation. However, most ML-based models extract few features from a single domain. Furthermore, studies have not examined effective electrode placement on skull, which affects process, while others employed feature selection approaches to reduce space dimension consequently complexity training models. This study presents an tool for automatically identifying ADHD entitled "ADHD-AID". present uses several multi-resolution analysis including variational mode decomposition, discrete wavelet transform, empirical decomposition. ADHD-AID extracts thirty time time-frequency domains identify ADHD, nonlinear features, band-power entropy-based statistical features. also looks at best EEG detecting ADHD. Additionally, it into location combinations that significant impact accuracy. variety methods choose those greatest influence diagnosis reducing classification's time. results show has provided scores accuracy, sensitivity, specificity, F1-score, Mathew correlation coefficients 0.991, 0.989, 0.992, 0.982, respectively, in with 10-fold cross-validation. Also, area under curve reached 0.9958. ADHD-AID's significantly higher than all earlier detection adolescents. These notable trustworthy findings support use such automated as means assistance doctors youngsters.

Язык: Английский

---