EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Impact of Simulated Radiation Dose Reduction on Deep Learning-Based Renal Segmentation Performance: A Simulation Study Using the KiTS21 (2021 Kidney and Kidney Tumor Segmentation Challenge) Dataset

Jae-Seoung Kim, Sung-Jong Eun

International Neurourology Journal · 31.05.2026 · 30 (Suppl 1) · S57-64

DOI: 10.5213/inj.2651124.062 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Neurourology Journal, erschienen am 31.05.2026. Verfasst von Jae-Seoung Kim und 1 weiteren Autor:innen. Beteiligte Einrichtung: Biomedical Research Center, Korea University Guro Hospital, Seoul, Korea. Die Arbeit nennt 19 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Purpose

This study aimed to quantitatively evaluate the effect of simulated radiation dose reduction on deep learning-based renal segmentation performance and to identify a clinically acceptable minimum dose threshold.

Methods

Using the KiTS21 (2021 Kidney and Kidney Tumor Segmentation Challenge) dataset, which included 299 contrastenhanced computed tomography volumes with expert segmentation labels, 4 dose levels were simulated: 100%, 50%, 25%, and 10%. Dose reduction was simulated using Poisson noise modeling. A 2-dimensional U-Net with a ResNet34 encoder was trained exclusively on standard-dose images and evaluated across all dose levels using 5-fold cross-validation. Case-level performance was assessed using the Dice similarity coefficient (DSC), intersection over union, 95th-percentile Hausdorff distance (HD95), and volumetric error. Statistical significance was evaluated using the Wilcoxon signed-rank test with effect-size analysis.

Results

At the standard dose, the model achieved a case-level DSC of 0.948±0.044. Performance remained stable at 50% dose (0.945±0.046), declined moderately at 25% dose (0.939±0.052), and decreased more substantially at 10% dose (0.921±0.069). The Wilcoxon signed-rank test showed statistically significant differences between 100% dose and all reduced dose levels (P0.93, down to 25% of the standard dose, suggesting that substantial dose reduction may be feasible without critically compromising artificial intelligence–assisted renal segmentation. The marked performance decline at 10% dose identifies a potential lower bound for clinical dose optimization.

Schlagwörter

Stichwörter der Autor:innen

Deep learningImage processing, computer-assistedKidneyRadiation dosageTomography, X-ray computed

Autor:innen und Institutionen

  • Biomedical Research Center, Korea University Guro Hospital, Seoul, Korea.
  • National IT Industry Promotion Agency, Jincheon, Korea. sjeun@nipa.kr.

Zitieren

APA

Kim, J. S., & Eun, S. J. (2026). Impact of Simulated Radiation Dose Reduction on Deep Learning-Based Renal Segmentation Performance: A Simulation Study Using the KiTS21 (2021 Kidney and Kidney Tumor Segmentation Challenge) Dataset. International Neurourology Journal, 30(Suppl 1), S57-64. https://doi.org/10.5213/inj.2651124.062

Vancouver

Kim JS, Eun SJ. Impact of Simulated Radiation Dose Reduction on Deep Learning-Based Renal Segmentation Performance: A Simulation Study Using the KiTS21 (2021 Kidney and Kidney Tumor Segmentation Challenge) Dataset. International Neurourology Journal. 2026;30(Suppl 1):S57-64. doi:10.5213/inj.2651124.062

BibTeX
@article{Kim2026,
  author = {Kim, Jae-Seoung and Eun, Sung-Jong},
  title = {Impact of Simulated Radiation Dose Reduction on Deep Learning-Based Renal Segmentation Performance: A Simulation Study Using the KiTS21 (2021 Kidney and Kidney Tumor Segmentation Challenge) Dataset},
  journal = {International Neurourology Journal},
  year = {2026},
  volume = {30},
  number = {Suppl 1},
  pages = {S57--64},
  doi = {10.5213/inj.2651124.062}
}
Kostenlos starten →