Preoperative Evaluation of Pituitary Macroadenomas Consistency Using Radiomic Features from Multi-Parametric MRI
Wan Tao1,2, Zhao Hui1,2, Li Deyu1,2, Ma Jun3, Wu Chunxue3, Meng Ming3, Qin Zengchang4*
1(School of Biomedical Science and Medical Engineering, Beihang University, Beijing 100191, China) 2(Beijing Advanced Innovation Center for Biomedical Engineering, Beihang University, Beijing 100191, China) 3(Department of Radiology, Beijing Tiantan Hospital, Capital Medical University, Beijing 100050, China) 4(School of Automation Science and Electrical Engineering, Beihang University, Beijing 100191, China)
Abstract：In order to explore the application value of image characteristics from multi-parameter magnetic resonance imaging (MP-MRI) in evaluating pituitary macroadenoma consistency, this paper presented a radiomics based computer-aided diagnosis method to accurately determine tumor consistency, thus providing an appropriate surgical approach. In this method, 6 types of texture features, a total number of 296, were extracted from tumor regions of MRIs (T1-weighted, T1-weighted contrast enhanced, T2-weighted). A feature selection method was adopted to identify important radiomic features. Two classifiers of support vector machine and random forest were utilized to distinguish soft and hard pituitary macroadenomas. The training, 10-fold cross validation and testing were performed on a total of 252 MRI images in 84 clinical studies. The experiment results showed that the feature combination of MP-MRI achieved better classification performance compared with single MRI protocol with classification accuracy, sensitivity, specificity and area under the curve of 89.80%, 90.51%, 89.88% and 94.08%, respectively. These suggested that MP-MRI features could effectively and accurately discriminate the soft from hard pituitary macroadenomas, which could be useful in improving the efficacy and prognosis of pituitary macroadenomas.
[1] Lopes M. The 2017 World Health Organization classification of tumors of the pituitary gland: A summary [J]. Acta Neuropathologica, 2017, 134(4): 521-535. [2] 武春雪,蒙茗,王政,等. 磁共振T2WI对垂体腺瘤质地的评估 [J]. 磁共振成像,2017,8(10):721-725. [3] Molitch M. Diagnosis and treatment of pituitary adenomas: A review [J]. The Journal of the American Medical Association (JAMA), 2017, 317(5): 516-524. [4] 王苗苗, 杨健. 垂体瘤 MRI 技术的研究进展[J]. 磁共振成像,2015(9):704-710. [5] Bahuleyan B, Raghuram L, Rajshekhar V, et al. To assess the ability of MRI to predict consistency of pituitary macroadenomas [J]. British Journal of Neurosurgery, 2006, 20(5): 324-326. [6] Yamamoto J, Kakeda S, Shimajiri S, et al. Tumor consistency of pituitary macroadenomas: predictive analysis on the basis of imaging features with contrast-enhanced 3D FIESTA at 3T [J]. NJNR American Journal of Neuroradiology, 2014, 35:297-303. [7] Mahmoud OM, Tominaga A, Amatya VJ, et al. Role of PROPELLER diffusion-weighted imaging and apparent diffusion coefficient in the evaluation of pituitary adenomas [J]. European Journal of Radiology, 2011, 80(2): 412-417. [8] 张利文,方梦捷,臧亚丽,等. 影像组学的发展与应用[J]. 中华放射学杂志,2017,51(1): 75-77. [9] Lambin P, Leijenaar RTH, Deist TM, et al. Radiomics: The bridge between medical imaging and personalized medicine [J]. Nature Reviews Clinical Oncology, 2017, 14: 749-762. [10] Zhang Xi, Lu Hongbing, Tian Qiang, et al. A radiomics nomogram based on multiparametric MRI might stratify glioblastoma patients according to survival [J]. European Radiology, 2019, 29(10): 5528-5538. [11] Vallieres M, Freeman CR, Skamene SR, et al. A radiomics model from joint FDG-PET and MRI texture features for the prediction of lung metastases in soft-tissue sarcomas of the extremities [J]. Physics in Medicine and Biology, 2015, 60(14): 5471-5496. [12] Wan Tao, Bloch BN, Plecha D, et al. A radio-genomics approach for identifying high risk estrogen receptor-positive breast cancers on DCE-MRI: Preliminary results in predicting OncotypeDX risk scores[J]. Scientific Reports, 2016, 6: 21394. [13] Wan Tao, Madabhushi A, Phinikaridou A, et al. Spatio-temporal texture (SpTeT) for distinguishing vulnerable from stable atherosclerotic plaque on dynamic contrast enhancement (DCE) MRI in a rabbit model [J]. Medical Physics, 2014, 41: 042303. [14] Collewet G, Strzelecki M, Mariette F. Influence of MRI acquisition protocols and image intensity normalization methods on texture classification [J]. Magnetic Resonance Imaging, 2004, 22(1): 81-91. [15] Zwanenburg A, Vallieres M, Abdalah MA, et al. The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput image-based phenotyping [J]. Radiology, 2020, 295(2): 328-338. [16] Hill P, Anantrasirichai N, Achim A, et al. Undecimated dual-tree complex wavelet transforms [J]. Signal Processing: Image Communication, 2015, 35: 61-70. [17] Wan Tao, Zhang Wanshu, Zhu Min, et al. Automated mitosis detection in histopathology based on non-gaussian modeling of complex wavelet coefficients[J]. Neurocomputing, 2017, 237: 291-303. [18] Peng Hanchuan, Long Fuhui, Ding C. Feature selection based on mutual information: criteria of max-dependency, max-relevance, and min-redundancy [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005, 27(8): 1226-1238. [19] Lu Yiping, Xiong Ji, Geng Daoying, et al. Prediction of the consistency of pituitary adenoma: A comparative study on diffusion-weighted imaging and pathological [J]. Journal of Neuroradiology, 2016, 43: 186-194. [20] Wei Liangfeng, Lin Shunan, Fan Kaichun, et al. Relationship between pituitary adenoma texture and collagen content revealed by comparative study of MRI and pathology analysis [J]. International Journal of Clinical and Experimental Medicine, 2015, 8(8): 12898-12905. [21] Wang Hui, Li Wensheng, Shi Dejin, et al. Expression of TGFbeta1 and pituitary adenoma fibrosis[J]. British Journal of Neurosurgery, 2009, 23(3): 293-296. [22] 杨晓莹,朱远,张敏,等. 垂体腺瘤质地与MRI图像的纹理分析比较[J]. 临床放射学杂志,2019,38(4): 613-617.
