End-to-End Deep-Learning-Based Diagnosis of Benign and Malignant Orbital Tumors on Computed Tomography Images - PubMed 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
. 2023 Jan 23;13(2):204.
doi: 10.3390/jpm13020204.

End-to-End Deep-Learning-Based Diagnosis of Benign and Malignant Orbital Tumors on Computed Tomography Images

Ji Shao  1 , Jiazhu Zhu  2 , Kai Jin  1 , Xiaojun Guan  3 , Tianming Jian  4 , Ying Xue  2 , Changjun Wang  1 , Xiaojun Xu  3 , Fengyuan Sun  4 , Ke Si  2 , Wei Gong  2 , Juan Ye  1
Affiliations

End-to-End Deep-Learning-Based Diagnosis of Benign and Malignant Orbital Tumors on Computed Tomography Images

Ji Shao et al. J Pers Med. .

Abstract

Determining the nature of orbital tumors is challenging for current imaging interpretation methods, which hinders timely treatment. This study aimed to propose an end-to-end deep learning system to automatically diagnose orbital tumors. A multi-center dataset of 602 non-contrast-enhanced computed tomography (CT) images were prepared. After image annotation and preprocessing, the CT images were used to train and test the deep learning (DL) model for the following two stages: orbital tumor segmentation and classification. The performance on the testing set was compared with the assessment of three ophthalmologists. For tumor segmentation, the model achieved a satisfactory performance, with an average dice similarity coefficient of 0.89. The classification model had an accuracy of 86.96%, a sensitivity of 80.00%, and a specificity of 94.12%. The area under the receiver operating characteristics curve (AUC) of the 10-fold cross-validation ranged from 0.8439 to 0.9546. There was no significant difference on diagnostic performance of the DL-based system and three ophthalmologists (p > 0.05). The proposed end-to-end deep learning system could deliver accurate segmentation and diagnosis of orbital tumors based on noninvasive CT images. Its effectiveness and independence from human interaction allow the potential for tumor screening in the orbit and other parts of the body.

Keywords: classification; deep learning; hemangioma; lymphoma; segmentation.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Workflow of the proposed method; CT, computed tomography.
Figure 2
Figure 2
Segmentation performance of the deep learning algorithm: (a,d) representative original computed tomography images from patients with hemangioma and lymphoma, respectively; (b,e) manual annotations which are shown in red; (c,f) automatic segmentation results which are shown in green; (g,h) recall-to-dice and precision-to-recall scatter diagrams.
Figure 3
Figure 3
Classification performance of deep-learning based model and 3 ophthalmologists: (a) the receiver operating characteristic (ROC) curve of ResNet-34; (b) result analysis of the 10-fold cross-validation; (c–f) confusion matrices of the deep-learning-based model and 3 ophthalmologists.
Figure 4
Figure 4
(a) Comparison of classification performance between the DL-based model and 3 ophthalmologists in the test set; (b) representative classification results; (c) comparison of the accuracy, sensitivity, and specificity of DL-based model and 3 ophthalmologists.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Shields J.A., Shields C.L., Scartozzi R. Survey of 1264 patients with orbital tumors and simulating lesions—The 2002 montgomery lecture, part 1. Ophthalmology. 2004;111:997–1008. doi: 10.1016/j.ophtha.2003.01.002. - DOI - PubMed
    1. Rootman D.B., Heran M.K., Rootman J., White V.A., Luemsamran P., Yucel Y.H. Cavernous venous malformations of the orbit (so-called cavernous haemangioma): A comprehensive evaluation of their clinical, imaging and histologic nature. Br. J. Ophthalmol. 2014;98:880–888. doi: 10.1136/bjophthalmol-2013-304460. - DOI - PubMed
    1. Scheuerle A.F., Steiner H.H., Kolling G., Kunze S., Aschoff A. Treatment and long-term outcome of patients with orbital cavernomas. Am. J. Ophthalmol. 2004;138:237–244. doi: 10.1016/j.ajo.2004.03.011. - DOI - PubMed
    1. Harris G.J. Cavernous hemangioma of the orbital apex: Pathogenetic considerations in surgical management. Am. J. Ophthalmol. 2010;150:764–773. doi: 10.1016/j.ajo.2010.07.027. - DOI - PubMed
    1. Low C.M., Stokken J.K. Typical orbital pathologies: Hemangioma. J. Neurol. Surg. B Skull Base. 2021;82:20–26. doi: 10.1055/s-0040-1722633. - DOI - PMC - PubMed

LinkOut - more resources