Weakly supervised machine learning

CAAI Transactions on Intelligence Technology, Год журнала: 2023, Номер 8(3), С. 549 - 580

Опубликована: Апрель 28, 2023

Abstract Supervised learning aims to build a function or model that seeks as many mappings possible between the training data and outputs, where each will predict label match its corresponding ground‐truth value. Although supervised has achieved great success in tasks, sufficient supervision for labels is not accessible domains because accurate labelling costly laborious, particularly medical image analysis. The cost of dataset with much higher than other domains. Therefore, it noteworthy focus on weakly analysis, more applicable practical applications. In this review, authors give an overview latest process including incomplete, inexact, inaccurate supervision, introduce related works different applications Related concepts are illustrated help readers get ranging from unsupervised within scope machine learning. Furthermore, challenges future analysis discussed.

Язык: Английский

---

Automated detection and forecasting of COVID-19 using deep learning techniques: A review

Neurocomputing, Год журнала: 2024, Номер 577, С. 127317 - 127317

Опубликована: Янв. 26, 2024

Язык: Английский

---

A review of deep learning in dentistry

Neurocomputing, Год журнала: 2023, Номер 554, С. 126629 - 126629

Опубликована: Июль 27, 2023

Oral diseases have a significant impact on human health, often going unnoticed in their early stages. Deep learning, promising field artificial intelligence, has shown remarkable success various domains, especially dentistry. This paper aims to provide an overview of recent research deep learning applications dentistry, with focus dental imaging. algorithms perform well difficult tasks such as image segmentation and recognition, enabling accurate identification oral conditions abnormalities. Integration other health data offers holistic understanding the relationship between systemic health. However, there are still many challenges that need be addressed.

Язык: Английский

---

Detection of Covid-19 and other pneumonia cases from CT and X-ray chest images using deep learning based on feature reuse residual block and depthwise dilated convolutions neural network

Applied Soft Computing, Год журнала: 2022, Номер 133, С. 109906 - 109906

Опубликована: Дек. 7, 2022

Язык: Английский

---

A Holistic Approach to Identify and Classify COVID-19 from Chest Radiographs, ECG, and CT-Scan Images Using ShuffleNet Convolutional Neural Network

Diagnostics, Год журнала: 2023, Номер 13(1), С. 162 - 162

Опубликована: Янв. 3, 2023

Early and precise COVID-19 identification analysis are pivotal in reducing the spread of COVID-19. Medical imaging techniques, such as chest X-ray or radiographs, computed tomography (CT) scan, electrocardiogram (ECG) trace images most widely known for early discovery coronavirus disease (COVID-19). Deep learning (DL) frameworks identifying positive patients literature limited to one data format, either ECG radiograph images. Moreover, using several types recover abnormal patterns caused by could potentially provide more information restrict virus. This study presents an effective detection classification approach Shufflenet CNN employing three images, i.e., radiograph, CT-scan, ECG-trace For this purpose, we performed extensive experiments with proposed each type image. With dataset, at different levels granularity, binary, three-class, four-class classifications. In addition, a binary experiment classifying CT-scan into COVID-positive normal. Finally, utilizing conducted five-class We evaluated baseline Radiography Database, SARS-CoV-2 dataset cardiac patients. The average accuracy 99.98% three-class scheme optimal 100% CT scans, 99.37% have proved efficacy our method over contemporary methods. scans gain 1.54% (in case images) from previous approach, which utilized first time, has major contribution improving prediction rate stages. Experimental findings demonstrate that framework outperforms models. example, state-of-the-art DL approaches, Squeezenet, Alexnet, Darknet19, achieving 99.98 (proposed method), 98.29, 98.50, 99.67, respectively.

Язык: Английский

---

Automated Classification of Brain Tumors from Magnetic Resonance Imaging Using Deep Learning

Brain Sciences, Год журнала: 2023, Номер 13(4), С. 602 - 602

Опубликована: Апрель 1, 2023

Brain tumor classification is crucial for medical evaluation in computer-assisted diagnostics (CAD). However, manual diagnosis of brain tumors from magnetic resonance imaging (MRI) can be time-consuming and complex, leading to inaccurate detection classification. This mainly because identification a complex procedure that relies on different modules. The advancements Deep Learning (DL) have assisted the automated process images various conditions, which benefits health sector. Convolutional Neural Network (CNN) one most prominent DL methods visual learning image tasks. study presents novel CNN algorithm classify types glioma, meningioma, pituitary. was tested benchmarked data compared with existing pre-trained VGG16, VGG19, ResNet50, MobileNetV2, InceptionV3 algorithms reported literature. experimental results indicated high accuracy 98.04%, precision, recall, f1-score success rate 98%, respectively. proved common kinds could categorized level accuracy. presented has good generalization capability execution speed helpful field medicine assist doctors making prompt accurate decisions associated diagnosis.

Язык: Английский

---

Dual-path information enhanced pyramid Unet for COVID-19 lung infection segmentation

Engineering Applications of Artificial Intelligence, Год журнала: 2025, Номер 142, С. 109977 - 109977

Опубликована: Янв. 5, 2025

Язык: Английский

---

A review on lung disease recognition by acoustic signal analysis with deep learning networks

Journal Of Big Data, Год журнала: 2023, Номер 10(1)

Опубликована: Июнь 12, 2023

Abstract Recently, assistive explanations for difficulties in the health check area have been made viable thanks considerable portion to technologies like deep learning and machine learning. Using auditory analysis medical imaging, they also increase predictive accuracy prompt early disease detection. Medical professionals are thankful such technological support since it helps them manage further patients because of shortage skilled human resources. In addition serious illnesses lung cancer respiratory diseases, plurality breathing is gradually rising endangering society. Because prediction immediate treatment crucial disorders, chest X-rays sound audio proving be quite helpful together. Compared related review studies on classification/detection using algorithms, only two based signal diagnosis conducted 2011 2018. This work provides a recognition with acoustic networks. We anticipate that physicians researchers working sound-signal-based will find this material beneficial.

Язык: Английский

---

Optimizing vitiligo diagnosis with ResNet and Swin transformer deep learning models: a study on performance and interpretability

Scientific Reports, Год журнала: 2024, Номер 14(1)

Опубликована: Апрель 21, 2024

Abstract Vitiligo is a hypopigmented skin disease characterized by the loss of melanin. The progressive nature and widespread incidence vitiligo necessitate timely accurate detection. Usually, single diagnostic test often falls short providing definitive confirmation condition, necessitating assessment dermatologists who specialize in vitiligo. However, current scarcity such specialized medical professionals presents significant challenge. To mitigate this issue enhance accuracy, it essential to build deep learning models that can support expedite detection process. This study endeavors establish framework accuracy end, comparative analysis five including ResNet (ResNet34, ResNet50, ResNet101 models) Swin Transformer series (Swin Base, Large models), were conducted under uniform condition identify model with superior classification capabilities. Moreover, sought augment interpretability these selecting one not only provides outcomes but also offers visual cues highlighting regions pertinent empirical findings reveal achieved best performance classification, whose AUC, sensitivity, specificity are 0.94, 93.82%, 94.02%, 93.5%, respectively. In terms interpretability, highlighted class activation map correspond lesion images, which shows effectively indicates specific category associated decision-making dermatological diagnosis. Additionally, visualization feature maps generated middle layer insights into internal mechanisms model, valuable for improving tuning performance, enhancing clinical applicability. underscore potential revolutionize diagnosis operational efficiency. research highlights necessity ongoing exploration domain fully leverage capabilities technologies diagnostics.

Язык: Английский

---

Artificial intelligence in pathologic diagnosis, prognosis and prediction of prostate cancer

American Journal of Clinical and Experimental Urology, Год журнала: 2024, Номер 12(4), С. 200 - 215

Опубликована: Янв. 1, 2024

Histopathology, which is the gold-standard for prostate cancer diagnosis, faces significant challenges. With ranking among most common cancers in United States and worldwide, pathologists experience an increased number biopsies. At same time, precise pathological assessment classification are necessary risk stratification treatment decisions care, adding to challenge pathologists. Recent advancement digital pathology makes artificial intelligence learning tools adopted histopathology feasible. In this review, we introduce concept of AI its various techniques field histopathology. We summarize clinical applications cancer, including grading, prognosis evaluation, options. also discuss how can be integrated into routine workflow. these rapid advancements, it evident that go beyond initial goal being diagnosis grading. Instead, provide additional information improve long-term patient outcomes by assessing detailed histopathologic features at pixel level using AI. Our review not only provides a comprehensive summary existing research but offers insights future advancements.

Язык: Английский

---