Integrative Approaches in Acute Ischemic Stroke: From Symptom Recognition to Future Innovations

Biomedicines, Journal Year: 2023, Volume and Issue: 11(10), P. 2617 - 2617

Published: Sept. 23, 2023

Among the high prevalence of cerebrovascular diseases nowadays, acute ischemic stroke stands out, representing a significant worldwide health issue with important socio-economic implications. Prompt diagnosis and intervention are milestones for management this multifaceted pathology, making understanding various stroke-onset symptoms crucial. A key role in is emphasizing essential multi-disciplinary team, therefore, increasing efficiency recognition treatment. Neuroimaging neuroradiology have evolved dramatically over years, multiple approaches that provide higher morphological aspects as well timely cerebral artery occlusions effective therapy planning. Regarding treatment matter, pharmacological approach, particularly fibrinolytic therapy, has its merits challenges. Endovascular thrombectomy, game-changer management, witnessed advances, technologies like stent retrievers aspiration catheters playing pivotal roles. For select patients, combining endovascular strategies offers evidence-backed benefits. The aim our comprehensive study on to efficiently compare current therapies, recognize novel possibilities from literature, describe state art interdisciplinary approach stroke. As we aspire holistic patient emphasis not just medical but also physical mental health, community engagement. future holds promising innovations, artificial intelligence poised reshape diagnostics treatments. Bridging gap between groundbreaking research clinical practice remains challenge, urging continuous collaboration research.

Language: Английский

---

Artificial intelligence for cardiovascular disease risk assessment in personalised framework: a scoping review

EClinicalMedicine, Journal Year: 2024, Volume and Issue: 73, P. 102660 - 102660

Published: May 27, 2024

Language: Английский

---

Deep learning models for tuberculosis detection and infected region visualization in chest X-ray images

Intelligent Medicine, Journal Year: 2023, Volume and Issue: 4(2), P. 104 - 113

Published: July 19, 2023

Tuberculosis is among the most frequent causes of infectious-disease-related mortality. Despite being treatable by antibiotics, tuberculosis often goes misdiagnosed and untreated, especially in rural low-resource areas. Chest X-rays are frequently used to aid diagnosis; however, this presents additional challenges because possibility abnormal radiological appearance a lack radiologists areas where infection prevalent. Implementing deep-learning-based imaging techniques for computer-aided diagnosis has potential enable accurate diagnoses lessen burden on medical specialists. In present work, we aimed develop segmentation classification models precise detection chest X-ray images, with visualization using Gradient-weighted Class Activation Mapping (Grad-CAM) heatmaps. First, trained UNet model 704 radiographs taken from Montgomery County Shenzhen Hospital datasets. Next, implemented 1400 control scans segment lung region. The images were NIAID TB portal program dataset. Then, applied deep learning Xception classify segmented region into normal classes. We further investigated capabilities Grad-CAM view abnormalities discuss them perspectives. For model, achieved accuracy, Jaccard index, Dice coefficient, area under curve values 96.35%, 90.38%, 94.88%, 0.99, respectively. precision, recall, F-1 score, 99.29%, 99.30%, 0.999, heatmap class showed similar patterns, lesions primarily upper part lungs. findings, including high accuracy detection, verify our system's efficacy superiority clinician precision raises valuable setup, particularly environments scarcity expertise.

Language: Английский

---

Machine Learning in Healthcare Analytics: A State-of-the-Art Review

Archives of Computational Methods in Engineering, Journal Year: 2024, Volume and Issue: unknown

Published: April 4, 2024

Language: Английский

---

Artificial intelligence in atherosclerotic disease: Applications and trends

Frontiers in Cardiovascular Medicine, Journal Year: 2023, Volume and Issue: 9

Published: Jan. 19, 2023

Atherosclerotic cardiovascular disease (ASCVD) is the most common cause of death globally. Increasing amounts highly diverse ASCVD data are becoming available and artificial intelligence (AI) techniques now bear promise utilizing them to improve diagnosis, advance understanding pathogenesis, enable outcome prediction, assist with clinical decision making promote precision medicine approaches. Machine learning (ML) algorithms in particular, already employed imaging applications facilitate automated detection experts believe that ML will transform field coming years. Current review first describes key concepts AI from a standpoint. We then provide focused overview current four main domains: coronary artery (CAD), peripheral arterial (PAD), abdominal aortic aneurysm (AAA), carotid disease. For each domain, presented refer primary modality used [e.g., computed tomography (CT) or invasive angiography] aim applied approaches, which include detection, phenotyping, assistance making. conclude strengths limitations future perspectives.

Language: Английский

---

Attention-Enabled Ensemble Deep Learning Models and Their Validation for Depression Detection: A Domain Adoption Paradigm

Diagnostics, Journal Year: 2023, Volume and Issue: 13(12), P. 2092 - 2092

Published: June 16, 2023

Depression is increasingly prevalent, leading to higher suicide risk. detection and sentimental analysis of text inputs in cross-domain frameworks are challenging. Solo deep learning (SDL) ensemble (EDL) models not robust enough. Recently, attention mechanisms have been introduced SDL. We hypothesize that attention-enabled EDL (aeEDL) architectures superior compared attention-not-enabled SDL (aneSDL) or aeSDL models. designed EDL-based with blocks build eleven kinds model five on four domain-specific datasets. scientifically validated our by comparing "seen" "unseen" paradigms (SUP). benchmarked results against the SemEval (2016) dataset established reliability tests. The mean increase accuracy for over their corresponding components was 4.49%. Regarding effect block, (AUC) aneSDL 2.58% (1.73%), aeEDL aneEDL 2.76% (2.80%). When vs. non-attention attention, greater than 4.82% (3.71%), 5.06% (4.81%). For benchmarking (SemEval), best-performing (ALBERT+BERT-BiLSTM) best (BERT-BiLSTM) 3.86%. Our scientific validation design showed a difference only 2.7% SUP, thereby meeting regulatory constraints. all hypotheses further demonstrated very effective generalized method detecting symptoms depression settings.

Language: Английский

---

Cardiovascular disease/stroke risk stratification in deep learning framework: a review

Cardiovascular Diagnosis and Therapy, Journal Year: 2023, Volume and Issue: 12(3), P. 557 - 598

Published: June 1, 2023

Abstract: The global mortality rate is known to be the highest due cardiovascular disease (CVD). Thus, preventive, and early CVD risk identification in a non-invasive manner vital as healthcare cost increasing day by day. Conventional methods for prediction of lack robustness non-linear relationship between factors events multi-ethnic cohorts. Few recently proposed machine learning-based stratification reviews without deep learning (DL) integration. study focuses on use techniques mainly solo (SDL) hybrid (HDL). Using PRISMA model, 286 DL-based studies were selected analyzed. databases included Science Direct, IEEE Xplore, PubMed, Google Scholar. This review focused different SDL HDL architectures, their characteristics, applications, scientific clinical validation, along with plaque tissue characterization CVD/stroke stratification. Since signal processing are also crucial, further briefly presented Electrocardiogram (ECG)-based solutions. Finally, bias AI systems. tools used (I) ranking method (RBS), (II) region-based map (RBM), (III) radial area (RBA), (IV) model assessment tool (PROBAST), (V) non-randomized studies-of interventions (ROBINS-I). surrogate carotid ultrasound image was mostly UNet-based DL framework arterial wall segmentation. Ground truth (GT) selection reducing (RoB) It observed that convolutional neural network (CNN) algorithms widely since feature extraction process automated. ensemble-based likely supersede paradigms. Due reliability, high accuracy, faster execution dedicated hardware, these powerful promising. can best reduced considering multicentre data collection evaluation.

Language: Английский

---

Research of segmentation recognition of small disease spots on apple leaves based on hybrid loss function and CBAM

Frontiers in Plant Science, Journal Year: 2023, Volume and Issue: 14

Published: June 6, 2023

Identification technology of apple diseases is great significance in improving production efficiency and quality. This paper has used Alternaria blotch brown spot disease leaves as the research object proposes a segmentation identification method based on DFL-UNet+CBAM to address problems low recognition accuracy poor performance small leaf recognition. The goal this accurately prevent control diseases, avoid fruit quality degradation yield reduction, reduce resulting economic losses. model employed hybrid loss function Dice Loss Focal added CBAM attention mechanism both effective feature layers extracted by backbone network results first upsampling, enhancing rescale inter-feature weighting relationships, enhance channel features spots suppressing healthy parts leaf, network’s ability extract while also increasing robustness. In general, after training, average rate improved decreases from 0.063 0.008 under premise ensuring image segmentation. smaller value is, better is. lesion test, MIoU was 91.07%, MPA 95.58%, F1 Score 95.16%, index increased 1.96%, predicted area actual overlap increased, 1.06%, category correctness 1.14%, number correctly identified pixels result more accurate. Specifically, compared original U-Net model, disease, 4.41%, 4.13%, Precision 1.49%, Recall 2.81%; spots, values 1.18%, 0.6%, 0.78%, 0.69%. diameter 0.2-0.3cm early stage, 0.5-0.6cm middle late stages, 0.3-3cm. Obviously, are larger than spots. noticeably, according quantitative analysis results, proving that model’s capacity segment greatly improved. findings demonstrate for detection suggested greater performance. can obtain sophisticated semantic information comparison traditional U-Net, further issues conventional methods well challenging convergence deep convolutional networks.

Language: Английский

---

Deep learning approach for cardiovascular disease risk stratification and survival analysis on a Canadian cohort

The International Journal of Cardiovascular Imaging, Journal Year: 2024, Volume and Issue: 40(6), P. 1283 - 1303

Published: April 28, 2024

Language: Английский

---

Atherosclerotic plaque classification in carotid ultrasound images using machine learning and explainable deep learning

Intelligent Medicine, Journal Year: 2023, Volume and Issue: 4(2), P. 83 - 95

Published: July 1, 2023

The incidence of cardiovascular diseases (CVD) is rising rapidly worldwide. Some forms CVD, such as stroke and heart attack, are more common among patients with certain conditions. Atherosclerosis development a major factor underlying events, attack stroke, its early detection may prevent events. Ultrasound imaging carotid arteries useful method for diagnosis atherosclerotic plaques; however, an automated to classify plaques evaluation early-stage CVD needed. Here, we propose classification high-risk plaque ultrasound images. Five deep learning (DL) models (VGG16, ResNet-50, GoogLeNet, XceptionNet, SqueezeNet) were used the results compared those machine (ML)-based technique, involving extraction 23 texture features from images using Support Vector Machine classifier. To enhance model interpretability, output gradient-weighted convolutional activation maps (gradCAMs) generated overlayed on original A series indices, including accuracy, sensitivity, specificity, F1-score, Cohen-kappa index, area under curve values, calculated evaluate performance. GradCAM allowed visualization most significant image regions. Of tested, GoogLeNet yielded highest accuracy (98.20%). ML also suitable applications requiring low computational resource, their (96.85%) similar that DL models. Further, can be completely than

Language: Английский

---