CT-based deep learning enables early postoperative recurrence prediction for intrahepatic cholangiocarcinoma - PubMed (original) (raw)

. 2022 May 19;12(1):8428.

doi: 10.1038/s41598-022-12604-8.

Keinosuke Ishido 2, Norihisa Kimura 2, Hayato Nagase 2, Taishu Kanda 2, Sotaro Ichiyama 3, Kenji Soma 3, Masashi Matsuzaka 4, Yoshihiro Sasaki 4, Shunsuke Kubota 2, Hiroaki Fujita 2, Takeyuki Sawano 5, Yutaka Umehara 5, Yusuke Wakasa 6, Yoshikazu Toyoki 6, Kenichi Hakamada 2

Affiliations

CT-based deep learning enables early postoperative recurrence prediction for intrahepatic cholangiocarcinoma

Taiichi Wakiya et al. Sci Rep. 2022.

Abstract

Preoperatively accurate evaluation of risk for early postoperative recurrence contributes to maximizing the therapeutic success for intrahepatic cholangiocarcinoma (iCCA) patients. This study aimed to investigate the potential of deep learning (DL) algorithms for predicting postoperative early recurrence through the use of preoperative images. We collected the dataset, including preoperative plain computed tomography (CT) images, from 41 patients undergoing curative surgery for iCCA at multiple institutions. We built a CT patch-based predictive model using a residual convolutional neural network and used fivefold cross-validation. The prediction accuracy of the model was analyzed. We defined early recurrence as recurrence within a year after surgical resection. Of the 41 patients, early recurrence was observed in 20 (48.8%). A total of 71,081 patches were extracted from the entire segmented tumor area of each patient. The average accuracy of the ResNet model for predicting early recurrence was 98.2% for the training dataset. In the validation dataset, the average sensitivity, specificity, and accuracy were 97.8%, 94.0%, and 96.5%, respectively. Furthermore, the area under the receiver operating characteristic curve was 0.994. Our CT-based DL model exhibited high predictive performance in projecting postoperative early recurrence, proposing a novel insight into iCCA management.

© 2022. The Author(s).

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1

Figure 1

The study workflow and methodological process.

Figure 2

Figure 2

The receiver operating characteristics curves of logistic regression analysis and the DL model. ROC curves show the performance of logistic regression analysis and the ResNet model in the validation dataset in detecting early recurrence. The AUC of logistic regression analysis is 0.770, and the average AUC of the convolutional neural network (CNN) model is 0.994.

Figure 3

Figure 3

A heatmap of iCCA on a preoperative plain CT using our prediction model. The color bar illustrates the degree of probability the model paid to it. Red areas represent a high risk of early recurrence; blue areas represent a low risk of early recurrence. (A) original image of a non-early recurrence case; (B) original image of an early recurrence case; (C) heatmap of a non-early recurrence case; (D) heatmap of an early recurrence case.

References

    1. Sung H, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: Cancer J. Clin. 2020 doi: 10.3322/caac.21660. - DOI - PubMed
    1. Aljiffry M, et al. Evidence-based approach to cholangiocarcinoma: A systematic review of the current literature. J. Am. Coll. Surg. 2009;208:134–147. doi: 10.1016/j.jamcollsurg.2008.09.007. - DOI - PubMed
    1. Bridgewater J, et al. Guidelines for the diagnosis and management of intrahepatic cholangiocarcinoma. J. Hepatol. 2014;60:1268–1289. doi: 10.1016/j.jhep.2014.01.021. - DOI - PubMed
    1. Society, A. C. Bile Duct Cancer Survival Rates|Cholangiocarcinoma Survival Rates, https://www.cancer.org/cancer/bile-duct-cancer/detection-diagnosis-stagi... (2021).
    1. Endo I, et al. Intrahepatic cholangiocarcinoma: Rising frequency, improved survival, and determinants of outcome after resection. Ann. Surg. 2008;248:84–96. doi: 10.1097/SLA.0b013e318176c4d3. - DOI - PubMed

MeSH terms

LinkOut - more resources