基于MR的垂体空细胞腺瘤与其他垂体腺瘤鉴别的影像组学研究
|
摘要
目的本文旨在通过垂体腺瘤的磁共振影像组学分析,对空细胞腺瘤及其他垂体腺瘤进行术前鉴别,进而为术前诊断、治疗及预后评估提供更多的参考信息。方法回顾性分析2013年1月至2018年5月吉林大学中日联谊医院182名垂体腺瘤患者临床及影像资料,所有患者术前均有MR平扫及增强图像以及术后免疫组化结果。对T1WI、T2WI平扫及T1WI增强序列的冠状位扫描图像进行手动分割并提取影像组学特征(每个序列图像分别提取1029个特征值),运用方差选择法、方差分析和最小绝对收缩和选择算法筛选出相对有效的影像组学特征。采用逻辑回归构建预测模型,预测表现通过受试者工作特征曲线(ROC曲线)评估,分析T1WI、T2WI及T1WI增强图像及图像组合在垂体免疫功能状态的预测能力,比较不同序列及组合对于垂体免疫学功能状态的优劣。结果训练集中,基于T1WI&T2WI序列图像的组合最终提取出13个有效的特征值,曲线下面积(AUC)值为0.80,95%置信区间(95%CI)为0.74～0.85,准确率为0.70。模型的预测效能在验证集中得到了检验,AUC值为0.70,95%CI为0.55～0.84,准确率为0.68。T1WI&T2WI&CE-T1WI序列组合几乎得到相同的结果。上述两组序列组合的模型效能比单个序列的模型预测效能更高。结论对影像组学为术前评估垂体腺瘤免疫组化结果提供了一种新的方式,为垂体空细胞腺瘤或其他类型垂体腺瘤的术前鉴别诊断、治疗方案的选择及预后评估提供一定参考价值。
Objective The article aimed to provide some information for the diagnosis, treatment and prognosis via histological analysis of pituitary adenoma by magnetic resonance imaging and discriminating the null cell adenoma and other pituitary adenomas before surgery. Methods The clinical and imaging data of 182 patients with pituitary adenoma in china-japan union hospital of Jilin university from January 2013 to May 2018 were retrospectively analyzed. All patients underwent MR plain scan, enhanced images and postoperative immunohistochemical examination. Coronal scanning images of T1WI, T2WI plain scan and T1WI enhanced sequence were manually segmented and radiomics features were extracted(1029 feature values were extracted for each sequence image). The effective radiomics features were selected via utilizing Variance threshold, Analysis of variance and the Least absolute shrinkage and selection operator ways. Logistic regression was used to construct the prediction model, and the prediction performance was evaluated by the ROC curve. The predictive ability of T1WI, T2WI and T1WI enhanced images and image combinations in the pituitary immune function state were analyzed, and the advantages and disadvantages of different sequences and combinations in the pituitary immune function state were compared. Results In the training cohort, 13 effective features extracted from T1WI and T2WI combined images were selected. The area under curve(AUC) value was 0.80, the 95% con?dence interval(95% CI) was 0.74～0.85,and the accuracy was 0.70. The ef?ciency of prediction model was validated in the validation cohort, with AUC of 0.70, 95% CI of 0.55～0.84 and accuracy of 0.68. Another model based on T1WI, T2WI and CE-T1WI got almost the same results. The model performance of the combination of the two sequences was higher than that of the single sequence. Conclusion Radiomics provides a new approach to discriminate the immunohistochemistry results of pituitary adenomas, which may helpful in the diagnosis, medical decision-making and prognosis for pituitary adenomas.
Objective The article aimed to provide some information for the diagnosis, treatment and prognosis via histological analysis of pituitary adenoma by magnetic resonance imaging and discriminating the null cell adenoma and other pituitary adenomas before surgery. Methods The clinical and imaging data of 182 patients with pituitary adenoma in china-japan union hospital of Jilin university from January 2013 to May 2018 were retrospectively analyzed. All patients underwent MR plain scan, enhanced images and postoperative immunohistochemical examination. Coronal scanning images of T1WI, T2WI plain scan and T1WI enhanced sequence were manually segmented and radiomics features were extracted(1029 feature values were extracted for each sequence image). The effective radiomics features were selected via utilizing Variance threshold, Analysis of variance and the Least absolute shrinkage and selection operator ways. Logistic regression was used to construct the prediction model, and the prediction performance was evaluated by the ROC curve. The predictive ability of T1WI, T2WI and T1WI enhanced images and image combinations in the pituitary immune function state were analyzed, and the advantages and disadvantages of different sequences and combinations in the pituitary immune function state were compared. Results In the training cohort, 13 effective features extracted from T1WI and T2WI combined images were selected. The area under curve(AUC) value was 0.80, the 95% con?dence interval(95% CI) was 0.74～0.85,and the accuracy was 0.70. The ef?ciency of prediction model was validated in the validation cohort, with AUC of 0.70, 95% CI of 0.55～0.84 and accuracy of 0.68. Another model based on T1WI, T2WI and CE-T1WI got almost the same results. The model performance of the combination of the two sequences was higher than that of the single sequence. Conclusion Radiomics provides a new approach to discriminate the immunohistochemistry results of pituitary adenomas, which may helpful in the diagnosis, medical decision-making and prognosis for pituitary adenomas.
引文
[1]Theodros D,Patel M,Ruzevick J,et al.Pituitary adenomas:historical perspective,surgical management and future directions[J].Cns Oncol,2015,4(6):411-429.
[2]Tjornstrand A,Gunnarsson K,Evert M,et al.The incidence rate of pituitary adenomas in western Sweden for the period 2001-2011[J].Eur J Endocrinol,2014,171(4):519-526.
[3]Lopes MBS.The 2017 World Health Organization classification of tumors of the pituitary gland:a summary[J].Acta Neuropathol,2017,134(4):521-535.
[4]Diri H,Ozaslan E,Kurtsoy A,et al.Is positive staining of non-functioning pituitary adenomas for luteinizing hormone associated with a poor prognosis?[J].Turk Neurosurg,2017,27(1):8-13.
[5]Roelfsema F,Biermasz NR,Pereira AM.Clinical factors involved in the recurrence of pituitary adenomas after surgical remission:a structured review and meta-analysis[J].Pituitary,2012,15(1):71-83.
[6]Neggers SJ,Aj VDL.Medical approach to pituitary tumors[J].Handb Clin Neurol,2014,124(124C):303.
[7]Molitch ME.Diagnosis and treatment of pituitary adenomas:Areview[J].Jama,2017,317(5):516-524.
[8]Tjornstrand A,Gunnarsson K,Evert M,et al.The incidence rate of pituitary adenomas in western Sweden for the period 2001-2011[J].Eur J Endocrinol,2014,171(4):519-526.
[9]Gillies RJ,Kinahan PE,Hricak H.Radiomics:Images are more than pictures,they are data[J].Radiology,2015,278(2):151169.
[10]Wu J,Aguilera T,Shultz D,et al.Early-stage non-small cell lung cancer:quantitative imaging characteristics of(18)F fluorodeoxyglucose PET/CT allow prediction of distant metastasis[J].Radiology,2016,281(1):151829.
[11]Li H,Zhu Y,Burnside ES,et al.MR imaging radiomics signatures for predicting the risk of breast cancer recurrence as given by research versions of mammaprint,oncotype DX,and PAM50 gene assays[J].Radiology,2017,281(2):152110.
[12]Huang YQ,Liang CH,He L,et al.Development and validation of a radiomics nomogram for preoperative prediction of lymph node metastasis in colorectal cancer[J].J Clin Oncol,2016,34(4):2157-2164.
[13]Oru?kaptan HH,Senmevsim?,?zcan OE,et al.Pituitary adenomas:results of 684 surgically treated patients and review of the literature[J].J Clin Oncol,2000,53(3):211-219.
[14]Molitch ME.Nonfunctioning pituitary tumors and pituitary incidentalomas[J].Endocrinol Metab Clin North Am,2008,37(1):151-171.
[15]Vilar L,Freitas MC,Naves LA,et al.Diagnosis and management of hyperprolactinemia:results of a Brazilian multicenter study with 1234 patients[J].J Endocrinol Invest,2008,31(5):436-444.
[16]Lake MG,Krook LS,Cruz SV.Pituitary adenomas:an overview[J].Am Fam Phys,2013,88(5):319-27.
[17]Mayson SE,Snyder PJ.Silent(clinically nonfunctioning pituitary adenomas[J].J Neurooncol,2014,117(3):429-436
[18]Mayson SE,Snyder PJ.Silent pituitary adenomas[J].Endocrin Metab Clin,2015,44(1):79-87.
[19]Zhang S,Song G,Zang Y,et al.Non-invasive radiomics approach potentially predicts non-functioning pituitary adenomas subtypes before surgery[J].Eur Radiol,2018:1-10.
[20]Galm BP,Martinez-Salazar EL,Swearingen B,et al.MRI texture analysis as a predictor of tumor recurrence or progression in patients with clinically non-functioning pituitary adenomas[J]Eur J Endocrinol,2018,179(3):191-198.
[21]Hossain MM,AlMuhanna K,Zhao L,et al.Semiautomatic segmentation of atherosclerotic carotid artery wall volume using 3D ultrasound imaging[J].Med Phys,2015,42(4):2029-2043.
[2]Tjornstrand A,Gunnarsson K,Evert M,et al.The incidence rate of pituitary adenomas in western Sweden for the period 2001-2011[J].Eur J Endocrinol,2014,171(4):519-526.
[3]Lopes MBS.The 2017 World Health Organization classification of tumors of the pituitary gland:a summary[J].Acta Neuropathol,2017,134(4):521-535.
[4]Diri H,Ozaslan E,Kurtsoy A,et al.Is positive staining of non-functioning pituitary adenomas for luteinizing hormone associated with a poor prognosis?[J].Turk Neurosurg,2017,27(1):8-13.
[5]Roelfsema F,Biermasz NR,Pereira AM.Clinical factors involved in the recurrence of pituitary adenomas after surgical remission:a structured review and meta-analysis[J].Pituitary,2012,15(1):71-83.
[6]Neggers SJ,Aj VDL.Medical approach to pituitary tumors[J].Handb Clin Neurol,2014,124(124C):303.
[7]Molitch ME.Diagnosis and treatment of pituitary adenomas:Areview[J].Jama,2017,317(5):516-524.
[8]Tjornstrand A,Gunnarsson K,Evert M,et al.The incidence rate of pituitary adenomas in western Sweden for the period 2001-2011[J].Eur J Endocrinol,2014,171(4):519-526.
[9]Gillies RJ,Kinahan PE,Hricak H.Radiomics:Images are more than pictures,they are data[J].Radiology,2015,278(2):151169.
[10]Wu J,Aguilera T,Shultz D,et al.Early-stage non-small cell lung cancer:quantitative imaging characteristics of(18)F fluorodeoxyglucose PET/CT allow prediction of distant metastasis[J].Radiology,2016,281(1):151829.
[11]Li H,Zhu Y,Burnside ES,et al.MR imaging radiomics signatures for predicting the risk of breast cancer recurrence as given by research versions of mammaprint,oncotype DX,and PAM50 gene assays[J].Radiology,2017,281(2):152110.
[12]Huang YQ,Liang CH,He L,et al.Development and validation of a radiomics nomogram for preoperative prediction of lymph node metastasis in colorectal cancer[J].J Clin Oncol,2016,34(4):2157-2164.
[13]Oru?kaptan HH,Senmevsim?,?zcan OE,et al.Pituitary adenomas:results of 684 surgically treated patients and review of the literature[J].J Clin Oncol,2000,53(3):211-219.
[14]Molitch ME.Nonfunctioning pituitary tumors and pituitary incidentalomas[J].Endocrinol Metab Clin North Am,2008,37(1):151-171.
[15]Vilar L,Freitas MC,Naves LA,et al.Diagnosis and management of hyperprolactinemia:results of a Brazilian multicenter study with 1234 patients[J].J Endocrinol Invest,2008,31(5):436-444.
[16]Lake MG,Krook LS,Cruz SV.Pituitary adenomas:an overview[J].Am Fam Phys,2013,88(5):319-27.
[17]Mayson SE,Snyder PJ.Silent(clinically nonfunctioning pituitary adenomas[J].J Neurooncol,2014,117(3):429-436
[18]Mayson SE,Snyder PJ.Silent pituitary adenomas[J].Endocrin Metab Clin,2015,44(1):79-87.
[19]Zhang S,Song G,Zang Y,et al.Non-invasive radiomics approach potentially predicts non-functioning pituitary adenomas subtypes before surgery[J].Eur Radiol,2018:1-10.
[20]Galm BP,Martinez-Salazar EL,Swearingen B,et al.MRI texture analysis as a predictor of tumor recurrence or progression in patients with clinically non-functioning pituitary adenomas[J]Eur J Endocrinol,2018,179(3):191-198.
[21]Hossain MM,AlMuhanna K,Zhao L,et al.Semiautomatic segmentation of atherosclerotic carotid artery wall volume using 3D ultrasound imaging[J].Med Phys,2015,42(4):2029-2043.