作者投稿
专家审稿
编辑办公
邮件订阅
RSS
Prediction of Survival Status of Locally Advanced Rectal Cancer after Neoadjuvant Chemoradiotherapy Based on Improved Fully Convolutional Networks Model
-
摘要:
目的 建立基于MRI影像图像预测局部进展期直肠癌(locally advanced rectal cancer，LARC)新辅助放化疗(neoadjuvant chemoradiotherapy，nCRT)后病理学完全缓解(pathological complete response，pCR)模型，以辅助患者个性化治疗方案的制订。 方法 回顾性纳入2013年6月至2018年12月中山大学附属第六医院接受nCRT治疗且行全直肠系膜切除术组织病理对治疗效果进行评定的LARC患者。按1:2的比例将患者依照住院时间先后顺序分为Data A与Data B 2个数据集。其中Data A数据集用于语义分割模型训练，Data B数据集按7:3的比例随机分为训练集和验证集，分别用于pCR预测模型训练与评价。收集Data A数据集病例的T2加权MRI影像资料，采用改进的全卷积网络(fully convolutional networks，FCN)模型对肿瘤区域进行语义分割，建立语义分割模型并提取最终卷积层中的影像特征。采用最小绝对值收缩和选择算子(least absolute shrinkage and selection operator, LASSO)回归法对提取的影像特征进行筛选，构建可预测pCR状态的支持向量机(support vector machine，SVM)分类器(预测模型)。以Data B训练集数据为基础，对该预测模型的性能进行训练，进一步在Data B验证集中对其性能进行评价。 结果 共入选符合纳入和排除标准的LARC患者304例，nCRT治疗后82例判定为pCR，222例为非pCR。2013年6月至2015年11月的103例患者为Data A数据集，2015年12月至2018年12月的201例患者为Data B数据集。Data B数据集中，训练集140例、验证集61例。改进的FCN模型对Data B数据集图像分割的Dice值为0.79 (95% CI: 0.65~0.81)，灵敏度为80%(95% CI: 77%~83%)，特异度为72%(95% CI: 64%~85%)。语义分割模型共提取最终卷积层中512个影像特征，经LASSO回归筛选后保留7个，用于pCR状态预测。预测模型在Data B训练集中预测pCR的曲线下面积(area under the curve, AUC)为0.65(95% CI: 0.61~0.71)，在Data B验证集中的AUC为0.69(95% CI: 0.59~0.74)。 结论 本研究提出的改进的FCN模型，对MRI图像进行语义分割具有较高的准确度。基于该方法构建的模型预测LARC患者接受nCRT治疗后pCR状态具有可行性。 Abstract:Objective To explore the model of using after neoadjuvant chemoradiotherapy (nCRT) imaging images of locally advanced rectal cancer (LARC) to predict whether pathological complete response (pCR) is achieved, thereby assisting physicians to develop a personalized plan. Methods Patients with LARC treated with nCRT at the Sixth Affiliated Hospital of Sun Yat-sen University from June 2013 to December 2018 were retrospectively included, and the treatment outcome was evaluated by total rectal mesenteric resection histopathology. The patients were divided into 2 data sets, Data A and Data B, according to the order of hospitalization time in a ratio of 1:2. Data A was used for semantic segmentation model training, and Data B was randomly divided into training and validation sets in the ratio of 7:3, which were used for pCR prediction model training and validation, respectively. The T2-weighted MRI images of Data A were collected, and the improved fully convolutional networks(FCN) model was used to semantically segment the tumor region, establish the semantic segmentation model and extract the image features in the final convolutional layer. The least absolute shrinkage and selection operator (LASSO) regression model was used to filter the extracted image features and construct a support vector machine (SVM) classifier that could predict the pCR state. The performance of the prediction model was trained on the basis of the Data B training set and further validated in the Data B validation set. Results A total of 304 patients with LARC who met the inclusion and exclusion criteria were enrolled, 82 patients reached pCR after nCRT, while 222 patients did not reach pCR (non-pCR). Among them, 103 patients from June 2013 to November 2015 were in Data A and 201 patients from December 2015 to December 2018 were in Data B. In Data B, 140 patients were in the training set and 61 patients in the validation set. The improved FCN model had a Dice value of 0.79(95% CI: 0.65-0.81), a sensitivity of 80%(95% CI: 77%-83%), and a specificity of 72%(95% CI: 64%-85%). A total of 512 image features in the final convolutional layer were extracted by the semantic segmentation model, and 7 were retained after LASSO regression screening for pCR state prediction. The area under the curve of the prediction model was 0.65(95% CI: 0.61-0.71) in the Data B training set and 0.69(95% CI: 0.59-0.74) in the Data B validation set for predicting pCR. Conclusions The improved FCN model proposed in this study has high accuracy for semantic segmentation of MRI images. The prediction model constructed based on this method is feasible to predict pCR status of LARC patients after receiving nCRT treatment. 作者贡献：王方负责思路框架设计与实现，并撰写论文初稿；庞晓琳负责数据收集、整理及伦理申请；范新娟提供研究思路，负责论文修订。利益冲突：所有作者均声明不存在利益冲突 -
表 1 304例LARC患者临床资料比较
指标 Data A(n=103) P值 Data B训练集(n=140) P值 Data B验证集(n=61) P值 pCR
(n=30)非pCR
(n=73)pCR
(n=37)非pCR
(n=103)pCR
(n=15)非pCR
(n=46)性别[n(%)] 0.831 0.133 0.150 男 8(26.7) 18(24.7) 22(59.5) 75(72.8) 7(46.7) 31(67.4) 女 22(73.3) 55(75.3) 15(40.5) 28(27.2) 8(53.3) 15(32.6) 年龄(x±s, 岁) 52.1±11.1 57.4±10.9 0.142 56.7±8.3 54.9±11.4 0.531 61.2±9.1 59.8±8.9 0.732 nCRT前T分期[n(%)] 0.983 0.214 0.031 T0 0(0) 0(0) 0(0) 0(0) 0(0) 0(0) T1 0(0) 0(0) 0(0) 0(0) 0(0) 0(0) T2 1(3.3) 3(4.1) 0(0) 4(3.9) 2(13.3) 0(0) T3 25(83.3) 60(82.2) 33(89.2) 79(76.7) 5(33.3) 13(28.3) T4 4(13.3) 10(13.7) 4(10.8) 20(19.4) 8(53.3) 33(71.7) nCRT前N分期[n(%)] 0.370 ＜0.010 0.060 N0 3(10.0) 13(17.8) 9(24.3) 30(29.1) 2(13.3) 2(4.3) N1 17(56.7) 31(42.5) 17(45.9) 21(20.4) 3(20.0) 2(4.3) N2 10(33.3) 29(39.7) 11(29.7) 52(50.5) 10(66.7) 42(91.3) nCRT后T分期[n(%)] ＜0.010 ＜0.010 ＜0.010 T0 30(100) 0(0) 20(54.1) 0(0) 15(100) 0(0) T1 0(0) 7(9.6) 17(45.9) 5(4.9) 0(0) 6(13.0) T2 0(0) 20(27.4) 0(0) 30(29.1) 0(0) 10(21.7) T3 0(0) 43(58.9) 0(0) 58(56.3) 0(0) 15(32.6) T4 0(0) 3(4.1) 0(0) 10(9.7) 0(0) 15(32.6) nCRT后N分期[n(%)] ＜0.010 ＜0.010 0.183 N0 30(100) 52(71.2) 37(100) 80(77.7) 15(100) 37(80.4) N1 0(0) 20(27.4) 0(0) 23(22.3) 0(0) 7(15.2) N2 0(0) 1(1.4) 0(0) 0(0) 0(0) 2(4.3) LARC：局部进展期直肠癌；nCRT: 新辅助放化疗; pCR: 病理学完全缓解 
下载: 导出CSV
-
[1] Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA Cancer J Clin, 2018, 68: 394-424. doi: 10.3322/caac.21492 [2] Burt RW, Barthel JS, Dunn KB, et al. NCCN clinical practice guidelines in oncology. Colorectal cancer screening[J]. J Natl Compr Canc Netw, 2010, 8: 8-61. doi: 10.6004/jnccn.2010.0003 [3] Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative chemoradiotherapy for rectal cancer[J]. N Engl J Med, 2004, 351: 1731-1740. doi: 10.1056/NEJMoa040694 [4] Beets-Tan RGH, Lambregts DMJ, Maas M, et al. Magnetic resonance imaging for clinical management of rectal cancer: updated recommendations from the 2016 European Society of Gastrointestinal and Abdominal Radiology (ESGAR) consensus meeting[J]. Eur Radiol, 2018, 28: 1465-1475. doi: 10.1007/s00330-017-5026-2 [5] Hupkens BJP, Martens MH, Stoot JH, et al. Quality of life in rectal cancer patients after chemoradiation: watch-and-wait policy versus standard resection-a matched-controlled study[J]. Dis Colon Rectum, 2017, 60: 1032-1040. doi: 10.1097/DCR.0000000000000862 [6] Cui Y, Yang X, Shi Z, et al. Radiomics analysis of multiparametric MRI for prediction of pathological complete response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer[J]. Eur Radiol, 2019, 29: 1211-1220. doi: 10.1007/s00330-018-5683-9 [7] Li Y, Liu W, Pei Q, et al. Predicting pathological complete response by comparing MRI-based radiomics pre-and postneoadjuvant radiotherapy for locally advanced rectal cancer[J]. Cancer Med, 2019, 8: 7244-7252. doi: 10.1002/cam4.2636 [8] Shi L, Zhang Y, Nie KE, et al. Machine learning for prediction of chemoradiation therapy response in rectal cancer using pre-treatment and mid-radiation multi-parametric MRI[J]. Magn Reson Imaging, 2019, 61: 33-40. doi: 10.1016/j.mri.2019.05.003 [9] Hu H, Jiang H, Wang S, et al. 3.0 T MRI IVIM-DWI for predicting the efficacy of neoadjuvant chemoradiation for locally advanced rectal cancer[J]. Abdom Radiol (NY), 2021, 46: 134-143. doi: 10.1007/s00261-020-02594-4 [10] Martí-Bonmatí L, Alberich-Bayarri A. Imaging biomarkers: development and clinical integration[M]. Cham: Springer, 2016: 1-376. [11] Lowekamp BC, Chen DT, Ibáñez L, et al. The design of SimpleITK[J]. Front Neuroinform, 2013, 7: 45. [12] 中华医学会放射学分会医学影像大数据与人工智能工作委员会, 中华医学会放射学分会腹部学组, 中华医学会放射学分会磁共振学组. 结直肠癌CT和MRI标注专家共识(2020)[J]. 中华放射学杂志, 2021, 55: 111-116. doi: 10.3760/cma.j.cn112149-20200706-00894 [13] Shin HC, Roth HR, Gao M, et al. Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning[J]. IEEE Trans Med Imaging, 2016, 35: 1285-1298. doi: 10.1109/TMI.2016.2528162 [14] Jiang X, Wang Y, Liu W, et al. Capsnet, cnn, fcn: Comparative performance evaluation for image classification[J]. Int J Mach Learn Comput, 2019, 9: 840-848. doi: 10.18178/ijmlc.2019.9.6.881 [15] Pramanik R, Bag S. Handwritten Bangla city name word recognition using CNN-based transfer learning and FCN[J]. Neural Comput Appl, 2021, 33: 9329-9341. doi: 10.1007/s00521-021-05693-5 [16] Sun S, Jiang B, Zheng Y, et al. A comparative study of CNN and FCN for histopathology whole slide image analysis[C]. International Conference on Image and Graphics, 2019: 558-567. [17] Wang F, Liu H. Understanding the behaviour of contras-tive loss[C]. Proceedings of the IEEE/CVF Confer-ence on Computer Vision and Pattern Recognition, 2021: 2495-2504. [18] Li X, Sun X, Meng Y, et al. Dice loss for data-imbalanced NLP tasks[J]. arXiv, 2019. https://doi.org/10.48550/arXiv.1911.02855. [19] Parikh R, Mathai A, Parikh S, et al. Understanding and using sensitivity, specificity and predictive values[J]. Indian J Ophthalmol, 2008, 56: 45-50. doi: 10.4103/0301-4738.37595 [20] Paszke A, Gross S, Massa F, et al. Pytorch: An imperative style, high-performance deep learning library[C]. Advances in Neural Information Processing Systems, 2019: 32. [21] 吴晓婷, 闫德勤. 数据降维方法分析与研究[J]. 計算機應用研究, 2009, 26: 2832-2835. https://cdmd.cnki.com.cn/Article/CDMD-10183-1014292110.htm [22] 黄铉. 特征降维技术的研究与进展[J]. 计算机科学, 2018, 45: 16-21. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJA2018S1004.htm [23] Ranstam J, Cook JA. LASSO regression[J]. J Br Surg, 2018, 105: 1348. doi: 10.1002/bjs.10895 [24] Chauhan VK, Dahiya K, Sharma A. Problem formulations and solvers in linear SVM: a review[J]. Artif Intell Rev, 2019, 52: 803-855. doi: 10.1007/s10462-018-9614-6 [25] Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn: Machine learning in python[J]. J Mach Learn Res, 2012 12: 2825-2830. [26] Khan Z, Yahya N, Alsaih K, et al. Evaluation of deep neural networks for semantic segmentation of prostate in T2W MRI[J]. Sensors(Basel), 2020, 20: 3183. [27] Ali HM, Kaiser MS, Mahmud M. Application of convolu-tional neural network in segmenting brain regions from MRI data[C]. International Conference on Brain Informatics, 2019: 136-146. [28] Long J, Shelhamer E, Darrell T. Fully convolutional net-works for semantic segmentation[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 3431-3440. [29] Huang CM, Huang MY, Huang CW, et al. Machine Learning for Predicting Pathological Complete Response in Patients with Locally Advanced Rectal Cancer after Neoadjuvant Chemoradiotherapy[J]. Sci Rep, 2020, 10: 12555. [30] Horvat N, Carlos Tavares Rocha C, Clemente Oliveira B, et al. Mri of rectal cancer: Tumor staging, imaging techniques, and management[J]. Radiographics, 2019, 39: 367-387. doi: 10.1148/rg.2019180114 [31] Wang H, Hu D, Yao H, et al. Radiomics analysis of multiparametric MRI for the preoperative evaluation of patholo-gical grade in bladder cancer tumors[J]. Eur Radiol, 2019, 29: 6182-6190. doi: 10.1007/s00330-019-06222-8 -
点击查看大图
图(6) / 表(1)
计量
- 文章访问数: 325
- HTML全文浏览量: 84
- PDF下载量: 52
- 被引次数: 0
