Robust Deep Learning Model for Black Fungus Detection Based on Gabor Filter and Transfer Learning

Computer Systems Science and Engineering, Journal Year: 2023, Volume and Issue: 47(2), P. 1507 - 1525

Published: Jan. 1, 2023

Black fungus is a rare and dangerous mycology that usually affects the brain lungs could be life-threatening in diabetic cases. Recently, some COVID-19 survivors, especially those with co-morbid diseases, have been susceptible to black fungus. Therefore, recovered patients should seek medical support when they notice mucormycosis symptoms. This paper proposes novel ensemble deep-learning model includes three pre-trained models: reset (50), VGG (19), Inception. Our approach medically intuitive efficient compared traditional deep learning models. An image dataset was aggregated from various resources divided into two classes: class skin infection class. To best of our knowledge, study first concerned building detection models based on algorithms. The proposed can significantly improve performance classification task increase generalization ability such binary task. According reported results, it has empirically achieved sensitivity value 0.9907, specificity 0.9938, precision negative predictive 0.9907.

Language: Английский

---

Genetic-efficient fine-tuning with layer pruning on multimodal Covid-19 medical imaging

Neural Computing and Applications, Journal Year: 2023, Volume and Issue: 36(6), P. 3215 - 3237

Published: Dec. 4, 2023

Abstract Medical image analysis using multiple modalities refers to the process of analyzing and extracting information from more than one type in order gain a comprehensive understanding given subject. To maximize potential multimodal data improving enhancing our disease, sophisticated classification techniques must be developed as part integration classify meaningful different types data. A pre-trained model, such those trained on large datasets ImageNet, has learned rich representations that can used for various downstream tasks. Fine-tuning model further developing knowledge gained pre-existing dataset. In comparison training scratch, fine-tuning allows transferred target task, thus performance efficiency. evolutionary search, genetic algorithm (GA) is an emulates natural selection genetics. this context, population candidate solutions generated, fitness evaluated new are generated by applying operations mutation crossover. Considering above characteristics, present study presents efficient architecture called Selective-COVIDNet COVID-19 cases novel selective layer-pruning algorithm. detect data, current will use fine-tune models adjusting specific layers selectively. Furthermore, proposed approach provides flexibility depth two deep learning architectures, VGG-16 MobileNet-V2. The impact freezing was assessed five strategies, namely Random, Odd, Even, Half, Full Freezing. Therefore, existing enhanced Covid-19 tasks while minimizing their computational burden. For evaluating effectiveness framework, multi-modal standard used, including CT-scan images electrocardiogram (ECG) recordings individuals with COVID-19. From conducted experiments, it found framework effectively accuracy 98.48% MobileNet-V2 99.65% VGG-16.

Language: Английский

---

Advanced hybrid attention-based deep learning network with heuristic algorithm for adaptive CT and PET image fusion in lung cancer detection

Medical Engineering & Physics, Journal Year: 2024, Volume and Issue: 126, P. 104138 - 104138

Published: March 4, 2024

Language: Английский

---

X-RCRNet: An explainable deep-learning network for COVID-19 detection using ECG beat signals

Biomedical Signal Processing and Control, Journal Year: 2023, Volume and Issue: 87, P. 105424 - 105424

Published: Sept. 19, 2023

Wearable systems measuring human physiological indicators with integrated sensors and supervised learning-based medical image analysis (e.g. ECG, X-ray, CT or ultrasound images for lung the chest) have been considered relevant tools COVID-19 monitoring diagnosis. However, these two technical roadmaps their respective advantages drawbacks. The current wearable enable to realize real-time of but are limited its basic symptoms only, neither allowing distinguish it from other diseases nor performing deep analysis. Current can provide accurate decision support diagnosis rarely deals data processing. In this context, we propose a new system by combining roadmaps. Considering that electrocardiogram (ECG) has proved evolution symptoms, proposed will integrate an explainable Deep Neural Network online gravity using ECG beat signal. This paper focus on model named X-RCRNet. network is based ResNet18 few enhancements: 1) LSTM Layers regenerating backpropagation error further extracting involved time-varying features; 2) LeakyReLU increasing performances model. With accuracy 96.48 % after experiments, our not only outperformed existing methods in terms robustness, also originally identify ST interval pattern, as most prominent key features affected virus.

Language: Английский

---

Longitudinal Chest X-ray Scores and their Relations with Clinical Variables and Outcomes in COVID-19 Patients

Diagnostics, Journal Year: 2023, Volume and Issue: 13(6), P. 1107 - 1107

Published: March 15, 2023

Background: This study evaluated the temporal characteristics of lung chest X-ray (CXR) scores in COVID-19 patients during hospitalization and how they relate to other clinical variables outcomes (alive or dead). Methods: is a retrospective patients. CXR disease severity were analyzed for: (i) survivors (N = 224) versus non-survivors 28) general floor group, (ii) 92) 56) invasive mechanical ventilation (IMV) group. Unpaired t-tests used compare between time points. Comparison across multiple points repeated measures ANOVA corrected for comparisons. Results: For general-floor patients, non-survivor significantly worse at admission compared those (p < 0.05), deteriorated outcome 0.05) whereas survivor did not > 0.05). IMV similar intubation both improved with showing greater improvement Hospitalization duration different groups correlated lactate dehydrogenase, respiratory rate, D-dimer, C-reactive protein, procalcitonin, ferritin, SpO2, lymphocyte count Conclusions: Longitudinal have potential provide prognosis, guide treatment, monitor progression.

Language: Английский

---