EUVIMEDEUVIMED entdecken
VergleichsstudieOpen Access · CC BY 4.0

Application of adaptive deep learning-based automatic segmentation in radiomics model for preoperative WHO/ISUP grading of clear cell renal cell carcinoma: a retrospective comparative study with manual segmentation

Hongqing Zhu, Zhihui Chen, Jianbo Zhang, Moran Yang, Kangchen Gu, Wenxia Bao, Yinlai Du, Sihui Hou, Wenjun Yao

PeerJ · 27.03.2026 · 14 · e21022

DOI: 10.7717/peerj.21022 ↗

Kurz gefasst

Vergleichsstudie in der Zeitschrift PeerJ, online erschienen am 27.03.2026. Verfasst von Hongqing Zhu und 8 weiteren Autor:innen. Beteiligte Einrichtung: The Second Affiliated Hospital, Anhui Medical University, Hefei, China. Hauptthemen laut MEDLINE: Carcinoma, Renal Cell, Kidney Neoplasms, Deep Learning. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University Natural Science Research Project of Anhui Province, Anhui Institute of Translational Medicine.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Aim

To evaluate the effectiveness of different methods for segmenting tumor regions of interest in building prediction models for the World Health Organization/International Society of Urological Pathology (WHO/ISUP) grade of clear cell renal cell carcinoma (ccRCC). Materials and Methods This retrospective single-center study analyzed computed tomography (CT) images of 405 patients (training/test cohort, 324/81) with pathologically confirmed ccRCC. Two methods were used for tumor segmentation: (1) automatic segmentation: the nnU-Net model, trained on the public KiTS19 dataset, and (2) manual segmentation. Radiomics features were extracted and selected from both automatically and manually segmented images. Support vector machine (SVM) and K-nearest neighbors were used to construct pathological grade prediction models. The segmentation accuracies of nnU-Net and manual annotation were compared. The receiver operating characteristic curve, area under the curve (AUC), accuracy, sensitivity, and specificity were used to evaluate diagnostic performance. The DeLong test was used to assess the differences between the models.

Results

The average Dice similarity coefficient was 0.842 ± 0.149. Automatic segmentation was time-efficient. A total of 1,834 features were extracted from each tumor. The AutoSeg-SVM model with nine features achieved the highest diagnostic performance, with an AUC value of 0.865 (0.726–1.000), an accuracy of 79.0%, sensitivity of 85.7%, and specificity of 77.6%. Both models in the automatic segmentation group showed comparable or slightly better performance than those in the manual segmentation group, although the differences were not statistically significant.

Conclusions

The nnU-Net automatic segmentation provided diagnostic efficacy comparable to manual segmentation in preoperative WHO/ISUP grade prediction for ccRCC. It significantly reduced the time required for lesion segmentation and improved workflow efficiency.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Carcinoma, Renal CellKidney NeoplasmsDeep LearningHumansRadiomicsRetrospective StudiesNeoplasm GradingFemaleMaleTomography, X-Ray ComputedMiddle AgedSupport Vector MachineAgedWorld Health Organization

Stichwörter der Autor:innen

Automatic segmentationClear cell renal cell carcinomaRadiomicsTumor grading predictionWorld Health Organization/International Society of Urological Pathology

Autor:innen und Institutionen

  • The Second Affiliated Hospital, Anhui Medical University, Hefei, China.
  • Medical Imaging Research Center, Anhui Medical University, Hefei, China.
  • Electronic Information Engineering, Anhui University, Hefei, China.
  • The Second Clinical Medical School, Anhui Medical University, Hefei, China.

Förderung

  • University Natural Science Research Project of Anhui Province
  • Anhui Institute of Translational Medicine

Zitieren

APA

Zhu, H., Chen, Z., Zhang, J., Yang, M., Gu, K., Bao, W., Du, Y., Hou, S., & Yao, W. (2026). Application of adaptive deep learning-based automatic segmentation in radiomics model for preoperative WHO/ISUP grading of clear cell renal cell carcinoma: a retrospective comparative study with manual segmentation. PeerJ, 14, e21022. https://doi.org/10.7717/peerj.21022

Vancouver

Zhu H, Chen Z, Zhang J, Yang M, Gu K, Bao W, et al. Application of adaptive deep learning-based automatic segmentation in radiomics model for preoperative WHO/ISUP grading of clear cell renal cell carcinoma: a retrospective comparative study with manual segmentation. PeerJ. 2026;14:e21022. doi:10.7717/peerj.21022

BibTeX
@article{Zhu2026,
  author = {Zhu, Hongqing and Chen, Zhihui and Zhang, Jianbo and Yang, Moran and Gu, Kangchen and Bao, Wenxia and Du, Yinlai and Hou, Sihui and Yao, Wenjun},
  title = {Application of adaptive deep learning-based automatic segmentation in radiomics model for preoperative WHO/ISUP grading of clear cell renal cell carcinoma: a retrospective comparative study with manual segmentation},
  journal = {PeerJ},
  year = {2026},
  volume = {14},
  pages = {e21022},
  doi = {10.7717/peerj.21022}
}
Kostenlos starten →