Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Aug:46:102034.
doi: 10.1016/j.tranon.2024.102034. Epub 2024 Jun 13.

Multiphase enhanced CT-based transformer for differential diagnosis and predicting surgical risk events of solid abdominal tumors in children

Yahui Han  1 , Biyun Li  2 , Xiaokun Yu  1 , Jianing Liu  1 , Wei Zhao  1 , Da Zhang  1 , Jiao Zhang  3
Affiliations

Multiphase enhanced CT-based transformer for differential diagnosis and predicting surgical risk events of solid abdominal tumors in children

Yahui Han et al. Transl Oncol. 2024 Aug.

Abstract

Background: For pediatric patients with solid abdominal tumors, early diagnosis can guide clinical treatment decisions, and comprehensive preoperative evaluation is essential to reduce surgical risk. The aim of this study was to explore the feasibility of multiphase enhanced CT-based transformer in the early diagnosis of tumors and prediction of surgical risk events (SRE).

Methods: A total of 496 pediatric patients with solid abdominal tumors were enrolled in the study. With Swin transformer, we constructed and trained two Swin-T models based on preoperative multiphase enhanced CT for personalized prediction of tumor type and SRE status. Subsequently, we comprehensively evaluated the performance of each model and constructed four benchmark models for performance comparison.

Results: There was no significant difference in SRE status between tumor types. In the diagnostic task, areas under the receiver operating characteristic curves (AUC) of the Swin-T model were 0.987 (95 % CI, 0.973-0.997) and 0.844 (95 % CI, 0.730-0.940) in the training and validation cohorts, respectively. In predicting SRE, AUCs of the Swin-T model were 0.920 (95 % CI, 0.885-0.948) and 0.741 (95 % CI, 0.632-0.838) in the training and test cohorts, respectively. The Swin-T model achieved the best performance in both classification tasks compared to benchmark models.

Conclusion: The Swin-T model is a promising tool to assist pediatricians in the differential diagnosis of abdominal tumors and in comprehensive preoperative evaluation.

Keywords: Classifier; Diagnosis; Multiphase enhanced CT; Swin transformer.

PubMed Disclaimer

Conflict of interest statement

Declaration of competing interest The authors declare no potential conflicts of interest to disclose.

Figures

Fig 1
Fig. 1
Flowchart of this study. NB: neuroblastoma, WT: Wilm's tumor, HB: hepatoblastoma.
Fig 2
Fig. 2
This study improved the Swin-T model architecture. This model accepts multiphase augmented CT as model input and the categories of the classification task as model output.
Fig 3
Fig. 3
Performance evaluation of the Swin-T model in two classification tasks. (A) The ROC curve for the training cohort of the Swin-T model in the diagnostic task. (B) The ROC curve for the validation cohort of the Swin-T model in the diagnostic task. (C) Comparison of the Swin-T model with four benchmark models in the diagnostic task. (D) The ROC curve for the training cohort of the Swin-T model in predicting SRE. (E) The ROC curve for the validation cohort of the Swin-T model in predicting SRE. (F) Comparison of the Swin-T model with four benchmark models in predicting SRE. (G) Additional accuracy evaluation indicators for the Swin-T model in two classification tasks.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Filbin M., Monje M. Developmental origins and emerging therapeutic opportunities for childhood cancer. Nat. Med. 2019;25(3):367–376. doi: 10.1038/s41591-019-0383-9. - DOI - PMC - PubMed
    1. Green D.M., D'Angio G.J., Beckwith J.B., Breslow N.E., Grundy P.E., Ritchey M.L., Thomas P.R. Wilms tumor. CA Cancer J. Clin. 1996;46(1):46–63. doi: 10.3322/canjclin.46.1.46. - DOI - PubMed
    1. Gatta G., Botta L., Rossi S., Aareleid T., Bielska-Lasota M., Clavel J., Dimitrova N., Jakab Z., Kaatsch P., Lacour B., Mallone S., Marcos-Gragera R., Minicozzi P., Sánchez-Pérez M.J., Sant M., Santaquilani M., Stiller C., Tavilla A., Trama A., Visser O., Peris-Bonet R. Childhood cancer survival in Europe 1999-2007: results of EUROCARE-5–a population-based study. Lancet Oncol. 2014;15(1):35–47. doi: 10.1016/s1470-2045(13)70548-5. - DOI - PubMed
    1. Lambin P., Leijenaar R.T.H., Deist T.M., Peerlings J., de Jong E.E.C., van Timmeren J., Sanduleanu S., Larue R., Even A.J.G., Jochems A., van Wijk Y., Woodruff H., van Soest J., Lustberg T., Roelofs E., van Elmpt W., Dekker A., Mottaghy F.M., Wildberger J.E., Walsh S. Radiomics: the bridge between medical imaging and personalized medicine. Nat. Rev. Clin. Oncol. 2017;14(12):749–762. doi: 10.1038/nrclinonc.2017.141. - DOI - PubMed
    1. Birkemeier K.L. Imaging of solid congenital abdominal masses: a review of the literature and practical approach to image interpretation. Pediatr. Radiol. 2020;50(13):1907–1920. doi: 10.1007/s00247-020-04678-1. - DOI - PubMed