腮腺為人體最大唾液線組織，透過磁共振影像可以呈現其樣貌與檢測其病理狀
況，本研究的目的在於建立針對磁共振影像的全自動識別腮腺腫瘤系統，透過
此系統之分析與判斷，檢測影像中的腮腺腫瘤，並將其分為三種類型，分別為
沃辛腫瘤、多腺形腫瘤與惡性腫瘤。在這項研究中，我們使用二維卷積神經網
絡和多模式的磁共振影像來進行腮腺腫瘤的分區和類型的判別。我們收集了多
種磁共振影像對比，分別為 T2、T1 以及擴散權重對比，並利用擴散權重影像
來計算擴散係數。我們設計了五種磁共振對比的組合，以比較各種影像組合對
於腮腺腫瘤的識別結果，透過比較使用磁共振影像的各種組合作為卷積神經網
路的輸入影像所獲得的結果，進而發現擴散相關參數有助於提高預測準確性。

The study presents an automatic segmentation and classification method for
detecting parotid gland tumors from MR images. Various MR imaging methods have
been shown their potential to detect the location of parotid gland tumors and categorize
them into three types, including Warthin tumor, pleomorphic adenoma, and malignant
tumor. The MR imaging methods included but not limited to T2-weighted, postcontrast
T1-weighted, and diffusion-weighted images. In this study, we used recently an
advanced convolution neural network and the multi-modality MRI images to conduct
the segmentation and classifications of parotid gland tumors. We used five
combinations of MRI contrasts as the input data of the neural network, and compared
the classification accuracy of parotid gland tumors. The results supported that diffusion-related parameters played the primary role of the prediction accuracy.

1. Guntinas-Lichius O, Gabriel B, Klussmann JP. Risk of facial palsy and severe
Frey's syndrome after conservative parotidectomy for benign disease:
analysis of 610 operations. Acta Otolaryngol 2006;126(10):1104-1109.
2. Koch M, Zenk J, Iro H. Long-term results of morbidity after parotid gland
surgery in benign disease. Laryngoscope 2010;120(4):724-730.
3. Lee YC, Park GC, Lee JW, Eun YG, Kim SW. Prevalence and risk factors of
sialocele formation after partial superficial parotidectomy: A multiinstitutional analysis of 357 consecutive patients. Head Neck 2016;38 Suppl
1:E941-944.
4. Jeong WJ, Park SJ, Cha W, Sung MW, Kim KH, Ahn SH. Fine needle
aspiration of parotid tumors: diagnostic utility from a clinical perspective. J
Oral Maxillofac Surg 2013;71(7):1278-1282.
5. Mallon DH, Kostalas M, MacPherson FJ, Parmar A, Drysdale A, Chisholm
E, Sadek S. The diagnostic value of fine needle aspiration in parotid lumps.
Ann R Coll Surg Engl 2013;95(4):258-262.
6. Chen X, Wei X, Zhang Z, Yang R, Zhu Y, Jiang X. Differentiation of trueprogression from pseudoprogression in glioblastoma treated with radiation
therapy and concomitant temozolomide by GLCM texture analysis of
conventional MRI. Clin Imaging 2015;39(5):775-780.
7. Kono K, Inoue Y, Nakayama K, Shakudo M, Morino M, Ohata K, Wakasa K,
Yamada R. The role of diffusion-weighted imaging in patients with brain
tumors. AJNR Am J Neuroradiol 2001;22(6):1081-1088.
8. Celebi I, Mahmutoglu AS, Ucgul A, Ulusay SM, Basak T, Basak M.
Quantitative diffusion-weighted magnetic resonance imaging in the
evaluation of parotid gland masses: a study with histopathological correlation.
Clin Imaging 2013;37(2):232-238.
9. Habermann CR, Arndt C, Graessner J, Diestel L, Petersen KU, Reitmeier F,
34
Ussmueller JO, Adam G, Jaehne M. Diffusion-weighted echo-planar MR
imaging of primary parotid gland tumors: is a prediction of different
histologic subtypes possible? AJNR Am J Neuroradiol 2009;30(3):591-596.
10. Habermann CR, Gossrau P, Graessner J, Arndt C, Cramer MC, Reitmeier F,
Jaehne M, Adam G. Diffusion-weighted echo-planar MRI: a valuable tool for
differentiating primary parotid gland tumors? Rofo 2005;177(7):940-945
11. Lechner Goyault J, Riehm S, Neuville A, Gentine A, Veillon F. Interest of
diffusion-weighted and gadolinium-enhanced dynamic MR sequences for the
diagnosis of parotid gland tumors. J Neuroradiol 2011;38(2):77-89.
12. Juan CJ, Chang HC, Hsueh CJ, Liu HS, Huang YC, Chung HW, Chen CY,
Kao HW, Huang GS. Salivary glands: echo-planar versus PROPELLER
Diffusion-weighted MR imaging for assessment of ADCs. Radiology
2009;253(1):144-152.
13. Shelhamer E, Long J, Darrell T. Fully Convolutional Networks for Semantic
Segmentation. IEEE Transactions on Pattern Analysis and Machine
Intelligence 2017;39(4):640-651.
14. Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for
biomedical image segmentation. 2015. Springer. p 234-241.
15. Pan SJ, Yang Q. A Survey on Transfer Learning. IEEE Transactions on
Knowledge and Data Engineering 2010;22(10):1345-1359.
16. Bakas S, Reyes M, Jakab A, Bauer S, Rempfler M, Crimi A, Takeshi
Shinohara R, Berger C, Ha SM, Rozycki M and others. Identifying the Best
Machine Learning Algorithms for Brain Tumor Segmentation, Progression
Assessment, and Overall Survival Prediction in the BRATS Challenge. eprint
arXiv:181102629 2018:arXiv:1811.02629.
17. Wang CW, Chu YH, Chiu DY, Shin N, Hsu HH, Lee JC, Juan CJ. JOURNAL
CLUB: The Warthin Tumor Score: A Simple and Reliable Method to
Distinguish Warthin Tumors From Pleomorphic Adenomas and Carcinomas.
AJR Am J Roentgenol 2018;210(6):1330-1337.
35
18. Freling NJ, Molenaar WM, Vermey A, Mooyaart EL, Panders AK, Annyas
AA, Thijn CJ. Malignant parotid tumors: clinical use of MR imaging and
histologic correlation. Radiology 1992;185(3):691-696.