EUVIMEDEUVIMED entdecken
Crossref

Effective deep learning classification for kidney stone using axial computed tomography (CT) images

Özlem Sabuncu iD, Bülent Bilgehan iD, Enver Kneebone, Omid Mirzaei

Biomedical Engineering / Biomedizinische Technik · 26.10.2023 · 68 (5) · 481-491

DOI: 10.1515/bmt-2022-0142 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Biomedical Engineering / Biomedizinische Technik, online erschienen am 03.05.2023. Verfasst von Özlem Sabuncu und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Electrical and Electronic Engineering, Near East University, Nicosia, Mersin, Türkiye. Hauptthemen laut MEDLINE: Deep Learning, Kidney Calculi. Die Arbeit wurde bisher 27-mal zitiert (Crossref) und nennt 43 Literaturangaben.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Stone formation in the kidneys is a common disease, and the high rate of recurrence and morbidity of the disease worries all patients with kidney stones. There are many imaging options for diagnosing and managing kidney stone disease, and CT imaging is the preferred method.

Objectives

Radiologists need to manually analyse large numbers of CT slices to diagnose kidney stones, and this process is laborious and time-consuming. This study used deep automated learning (DL) algorithms to analyse kidney stones. The primary purpose of this study is to classify kidney stones accurately from CT scans using deep learning algorithms.

Methods

The Inception-V3 model was selected as a reference in this study. Pre-trained with other CNN architectures were applied to a recorded dataset of abdominal CT scans of patients with kidney stones labelled by a radiologist. The minibatch size has been modified to 7, and the initial learning rate was 0.0085.

Results

The performance of the eight models has been analysed with 8209 CT images recorded at the hospital for the first time. The training and test phases were processed with limited authentic recorded CT images. The outcome result of the test shows that the Inception-V3 model has a test accuracy of 98.52 % using CT images in detecting kidney stones.

Conclusions

The observation is that the Inception-V3 model is successful in detecting kidney stones of small size. The performance of the Inception-V3 Model is at a high level and can be used for clinical applications. The research helps the radiologist identify kidney stones with less computational cost and disregards the need for many experts for such applications.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningKidney CalculiHumansTomography, X-Ray ComputedAlgorithms

Stichwörter der Autor:innen

artificial intelligencecomputed tomographyconvolutional neural networkdeep learningkidney stone

Autor:innen und Institutionen

  • Department of Electrical and Electronic Engineering, Near East University, Nicosia, Mersin, Türkiye.
  • Vocational School of Health Services, European University of Lefke, Lefke, Mersin, Türkiye.
  • Department of Biomedical Engineering, Near East University, Nicosia, Mersin, Türkiye.

ORCID-Profile: Özlem Sabuncu, Bülent Bilgehan

Zitieren

APA

Sabuncu, Ö., Bilgehan, B., Kneebone, E., & Mirzaei, O. (2023). Effective deep learning classification for kidney stone using axial computed tomography (CT) images. Biomedical Engineering / Biomedizinische Technik, 68(5), 481-491. https://doi.org/10.1515/bmt-2022-0142

Vancouver

Sabuncu Ö, Bilgehan B, Kneebone E, Mirzaei O. Effective deep learning classification for kidney stone using axial computed tomography (CT) images. Biomedical Engineering / Biomedizinische Technik. 2023;68(5):481-491. doi:10.1515/bmt-2022-0142

BibTeX
@article{Sabuncu2023,
  author = {Sabuncu, Özlem and Bilgehan, Bülent and Kneebone, Enver and Mirzaei, Omid},
  title = {Effective deep learning classification for kidney stone using axial computed tomography (CT) images},
  journal = {Biomedical Engineering / Biomedizinische Technik},
  year = {2023},
  volume = {68},
  number = {5},
  pages = {481--491},
  doi = {10.1515/bmt-2022-0142}
}
Kostenlos starten →