Initial Analysis of Multiple Retinal Diseases Classification with Fuzzy Medical Image Retrieval

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27240%2F23%3A10254717" target="_blank" >RIV/61989100:27240/23:10254717 - isvavai.cz</a>

Výsledek na webu

<a href="https://ieeexplore.ieee.org/document/10394609" target="_blank" >https://ieeexplore.ieee.org/document/10394609</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/SMC53992.2023.10394609" target="_blank" >10.1109/SMC53992.2023.10394609</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Initial Analysis of Multiple Retinal Diseases Classification with Fuzzy Medical Image Retrieval

Popis výsledku v původním jazyce

Medical image retrieval is a highly discussed topic, and it includes an efficient classification of diagnoses based on the similarity search in large databases of medical images. It is very important for early and correct diagnosis and treatment. In this paper, we focus on detecting four diagnoses of treatable retinal diseases in optical coherence tomography (OCT) images. The fuzzy medical image retrieval model (FMIR) is applied to transfer images to fuzzy signatures organized in Fuzzy S-tree, as it was previously successfully used for breast cancer detection and COVID-19 chest X-ray detection. The paper examines and compares the performance of the FMIR method on 4-class and binary classification models built on an OCT dataset and compares the impact of two metrics, Euclidean and Hamming fuzzy distances. The experiments show a clear dominance of Hamming fuzzy distance. The best accuracy is achieved for binary classification (61.16 - 93.8%), while the performance of the 4-class model is worse (51.7%). The distribution of signature space and classification performance are analyzed in detail.

Název v anglickém jazyce

Initial Analysis of Multiple Retinal Diseases Classification with Fuzzy Medical Image Retrieval

Popis výsledku anglicky

Medical image retrieval is a highly discussed topic, and it includes an efficient classification of diagnoses based on the similarity search in large databases of medical images. It is very important for early and correct diagnosis and treatment. In this paper, we focus on detecting four diagnoses of treatable retinal diseases in optical coherence tomography (OCT) images. The fuzzy medical image retrieval model (FMIR) is applied to transfer images to fuzzy signatures organized in Fuzzy S-tree, as it was previously successfully used for breast cancer detection and COVID-19 chest X-ray detection. The paper examines and compares the performance of the FMIR method on 4-class and binary classification models built on an OCT dataset and compares the impact of two metrics, Euclidean and Hamming fuzzy distances. The experiments show a clear dominance of Hamming fuzzy distance. The best accuracy is achieved for binary classification (61.16 - 93.8%), while the performance of the 4-class model is worse (51.7%). The distribution of signature space and classification performance are analyzed in detail.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10200 - Computer and information sciences

Projekt

<a href="/cs/project/GF22-34873K" target="_blank" >GF22-34873K: Vícekriteriální optimalizace s omezeními pomocí analýzy potenciálních ploch</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics 2023

ISBN

979-8-3503-3703-7

ISSN

1062-922X

e-ISSN

2577-1655

Počet stran výsledku

6

Strana od-do

4926-4931

Název nakladatele

IEEE

Místo vydání

Piscataway

Místo konání akce

Honolulu

Datum konání akce

1. 10. 2023

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—