A Comparative Study on Multi-modal Fusion for Automated Lung Disease Diagnostics DOI
Sachin Kumar, Pradeep Kumar Mallick,

Olga Vorfolomeyeva

и другие.

Опубликована: Фев. 9, 2024

High-quality X-rays are now available to diagnose lung diseases with the help of radiologists. However, diagnostic process is time consuming and depends on specialist availability in medical institutions. Patient information may include chest varying quality, test results, doctors' notes prescriptions, medication details, among others. In this study, we present a model for classifying pulmonary using multimodal data from patient clinical studies radiographic images. Various methods were used generate artificial samples both images tabular laboratory study results during preparation. We also proposed method establishing correspondence between generated modals. The late fusion architecture was implemented. conducted experiments data-set two modalities. Results shows that an increase accuracy other parameters observed our comparison image only modality modality. It strengthen fact multimodality provides more insight learn provide precise diagnosis than single

Язык: Английский

Dynamic data‐driven railway bridge construction knowledge graph update method DOI
Jianbo Lai, Jun Zhu, Yukun Guo

и другие.

Transactions in GIS, Год журнала: 2023, Номер 27(7), С. 2099 - 2117

Опубликована: Окт. 20, 2023

Abstract Effectively integrating and correlating multisource data involved in the bridge construction process is crucial for improvement of informatization level. In current issues dynamic numerous low information sharing between different engineering departments, traditional management methods are inefficient providing comprehensive accurate support safety. Focusing on stage, this article proposes a data‐driven method railway knowledge graph (KG) combination with (materials, personnel, equipment sensors) KG technology. By taking as case, study develops prototype system analyzes effectiveness material traceability, personnel safety guidance, which can provide optimization. The results show that: (1) that takes into account features projects effectively integrate multiple elements; (2) dynamically updated through real‐time comparison advance prediction based collected by multi‐sensing at site, effective guiding safety; (3) assisted risk event decision‐making. comparative experiment group spreadsheet showed utilizing saved approximately 50% time achieved 20% higher accuracy rate traceability task compared to group. general, KG, realize integration process, necessary scientific basis fine management, help improve

Язык: Английский

Процитировано

8
Cross-modal Deep Learning-based Clinical Recommendation System for Radiology Report Generation from Chest X-rays DOI Open Access
Shashank Shetty,

V. S. Ananthanarayana,

Ajit Mahale

и другие.

International journal of engineering. Transactions B: Applications, Год журнала: 2023, Номер 36(8), С. 1569 - 1577

Опубликована: Янв. 1, 2023

Radiology report generation is a critical task for radiologists, and automating the process can significantly simplify their workload. However, creating accurate reliable radiology reports requires radiologists to have sufficient experience time review medical images. Unfortunately, many end with ambiguous conclusions, resulting in additional testing diagnostic procedures patients. To address this, we proposed an encoder-decoder-based deep learning framework that utilizes chest X-ray images produce reports. In our study, introduced novel text modelling visual feature extraction strategy as part of framework. Our approach aims extract essential textual information from generate more Additionally, developed dynamic web portal accepts X-rays input generates output. We conducted extensive analysis model compared its performance other state-of-the-art approaches. findings indicate significant improvement achieved by existing models, evidenced higher BLEU scores (BLEU1 = 0.588, BLEU2 0.4325, BLEU3 0.4017, BLEU4 0.3860) attained on Indiana University Dataset. These results underscore potential enhance accuracy reliability reports, leading efficient effective treatment.

Язык: Английский

Процитировано

7
Effect of Multimodal Metadata Augmentation on Classification Performance in Deep Learning DOI
Yuri Gordienko, Maksym Shulha, Sergii Stirenko

и другие.

Algorithms for intelligent systems, Год журнала: 2024, Номер unknown, С. 391 - 405

Опубликована: Янв. 1, 2024

Язык: Английский

Процитировано

1
SMC-CNN: Stacked Multi-Channel Convolution Neural Network for predicting Acute Brain Infarct from Magnetic Resonance Imaging Sequences DOI Creative Commons
Shashank Shetty,

V. S. Ananthanarayana,

Ajit Mahale

и другие.

IEEE Access, Год журнала: 2024, Номер 12, С. 171112 - 171142

Опубликована: Янв. 1, 2024

Язык: Английский

Процитировано

1
An Intelligent and Secure Real-Time Environment Monitoring System for Healthcare Using IoT and Cloud Computing with the Mobile Application Support DOI
Shashank Shetty

Springer eBooks, Год журнала: 2023, Номер unknown, С. 83 - 95

Опубликована: Янв. 1, 2023

Язык: Английский

Процитировано

3
Diagnostic Performance Evaluation of Deep Learning-Based Medical Text Modelling to Predict Pulmonary Diseases from Unstructured Radiology Free-Text Reports DOI Creative Commons
Shashank Shetty,

V. S. Ananthanarayana,

Ajit Mahale

и другие.

Acta Informatica Pragensia, Год журнала: 2023, Номер 12(2), С. 260 - 274

Опубликована: Сен. 5, 2023

The third most common cause of death worldwide is attributed to pulmonary diseases, making it imperative diagnose them promptly.Radiology a medical discipline that utilizes imaging guide treatment.Radiologists prepare reports interpreting the details and findings analyzed from images.Radiology free-text contain rich source textual information can be exploited enhance efficacy prognosis, treatment research.The radiology exist in an unstructured format as not conducive by themselves applied mathematical computation or Machine learning operations.Therefore, Natural Language Processing (NLP) strategies are employed convert natural language text into structured ingested Learning (ML) Deep (DL) models for extraction.We propose DL-based modelling framework incorporating knowledge base predict diseases reports.We have performed detailed diagnostic performance evaluations our proposed technique comparing with state-ofthe-art NLP techniques on re-ports extracted two institutions.The comprehensive analysis shows model has achieved superior results compared existing state-of-the-art techniques.

Язык: Английский

Процитировано

1
Data Augmentation vs. Synthetic Data Generation: An Empirical Evaluation for Enhancing Radiology Image Classification DOI
Shashank Shetty,

Ananthanarayana V.S.,

Ajit Mahale

и другие.

Опубликована: Авг. 25, 2023

Radiology is a field of medicine dealing with diagnostic images to detect diseases for further treatment. Collecting and annotating like Magnetic Resonance Imaging (MRI) X-Ray rigorous time-consuming process. Deep Learning methods are widely utilized disease classification prediction from images, but they demand substantial amounts training data. Additionally, certain uncommon in large patient cohorts, posing difficulties obtaining sufficient imaging samples construct accurate deep learning models. Data augmentation techniques commonly used overcome this challenge limited These involve applying geometric transformations such as rotation, cropping, zooming, flipping, other similar operations the enlarge dataset artificially. Another possible way expanding by synthesizing data generate artificial medical mimicking original images. This study presents RAD-DCGAN: A Convolutional Generative Adversarial Network produce high-resolution synthetic radiology X-ray MRI collected private hospital (KMC Hospital, India). We aim determine most effective technique enhancing performance image classifiers comparing evaluating proposed RAD-DCGAN standard strategy. Our empirical evaluation, which involved eight models, demonstrated that trained on outperformed those augmented The utilization model testing models has notable improvement 4-5% accuracy compared conventional techniques. signifies state-of-the-art achieved

Язык: Английский

Процитировано

1
A Comparative Study on Multi-modal Fusion for Automated Lung Disease Diagnostics DOI
Sachin Kumar, Pradeep Kumar Mallick,

Olga Vorfolomeyeva

и другие.

Опубликована: Фев. 9, 2024

High-quality X-rays are now available to diagnose lung diseases with the help of radiologists. However, diagnostic process is time consuming and depends on specialist availability in medical institutions. Patient information may include chest varying quality, test results, doctors' notes prescriptions, medication details, among others. In this study, we present a model for classifying pulmonary using multimodal data from patient clinical studies radiographic images. Various methods were used generate artificial samples both images tabular laboratory study results during preparation. We also proposed method establishing correspondence between generated modals. The late fusion architecture was implemented. conducted experiments data-set two modalities. Results shows that an increase accuracy other parameters observed our comparison image only modality modality. It strengthen fact multimodality provides more insight learn provide precise diagnosis than single

Язык: Английский

Процитировано

0