Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ Адаптивні системи ав...arrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
addClaim

This Research product is the result of merged Research products in OpenAIRE.

You have already added 0 works in your ORCID record related to the merged Research product.

Application of Transfer Learning for enhanced pulmonary disease detection via CT image embeddings

Application of Transfer Learning for enhanced pulmonary disease detection via CT image embeddings

Abstract

This study explores the use of transfer learning in enhancing pulmonary disease detection from computed tomography (CT) scans, focusing on the adaptation of a convolutional neural network (CNN) for generating CT slice embeddings. Dataset includes CT images of COVID-19 patients, healthy individuals, and those with community-acquired pneumonia. A multi-class classification network from our prior research was augmented to produce embedding vectors of given CT slices. Transfer learning played a pivotal role, wherein we reused and froze the convolutional layers of the original network, directing our training efforts on the new embedding output layer. To ensure the production of distinctive vectors, triplet loss was employed. Embeddings produced with acquired network were then utilized to train a KNeighborsClassifier, which achieved an accuracy of 0.987 in multi-class classification. This performance, supported by a detailed confusion matrix, signifies the effectiveness of our approach in medical diagnostics. Achieved results, while preliminary, demonstrate the potential of embedding-based classification systems in CT scan analysis, especially for COVID-19 diagnostics. Ref. 9, pic. 2

У даному досліджені розглянуто використання передавального навчання для покращення виявлення захворювань легень за допомогою комп'ютерної томографії (КТ). Розроблену у попередньому досліджені згорткову нейронну мережудля мультикласової класифікації було доповнено та дотреновано для створення вкладених представлень КТ-зрізів. Згорткові шари нейронної мережі було заморожено, а решту – замінено рядом шарів для виводу векторного представлення знімку, які і були дотреновані у даному дослідженні. Для забезпечення створення розрізнених векторівбуло обрано використати triplet loss. Вкладені представлення, створені з отриманою мережею, було використано для навчання класифікатора KNeighborsClassifier, який досяг точності 0.987 у багатокласовій класифікації. Отримані результати, хоча й є попередніми, демонструють потенціал систем класифікації на основі вкладених представлень у аналізі КТ-сканування, особливо для діагностики COVID-19. Бібл. 9, іл. 2

Keywords

embedding, computed tomography scan analysis, convolutional neural networks, аналіз знімків компʼютерної томографії, передавальне навчання, COVID-19, мультикласова класифікація, multiclass classification, transfer learning, згорткові нейронні мережі, вкладене представлення

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
gold
Related to Research communities