Prediction of High‐Risk Cytogenetic Status in Multiple Myeloma Based on Magnetic Resonance Imaging: Utility of Radiomics and Comparison of Machine Learning Methods

Journal of Magnetic Resonance Imaging - Tập 54 Số 4 - Trang 1303-1311 - 2021
Jianfang Liu1, Piaoe Zeng1, Wei Guo1, Chunjie Wang1, Yayuan Geng2, Ning Lang1, Huishu Yuan1
1Department of Radiology, Peking University Third Hospital, Haidian District, Beijing, People’s Republic of China
2Huiying Medical Technology (Beijing) Co., Ltd, Beijing, China

Tóm tắt

BackgroundRadiomics has shown promising results in the diagnosis, efficacy, and prognostic assessments of multiple myeloma (MM). However, little evidence exists on the utility of radiomics in predicting a high‐risk cytogenetic (HRC) status in MM.PurposeTo develop and test a magnetic resonance imaging (MRI)‐based radiomics model for predicting an HRC status in MM patients.Study TypeRetrospective.PopulationEighty‐nine MM patients (HRC [n: 37] and non‐HRC [n: 52]).Field Strength/SequenceA 3.0 T; fast spin‐echo (FSE): T1‐weighted image (T1WI) and fat‐suppression T2WI (FS‐T2WI).AssessmentOverall, 1409 radiomics features were extracted from each volume of interest drawn by radiologists. Three sequential feature selection steps—variance threshold, SelectKBest, and least absolute shrinkage selection operator—were repeated 10 times with 5‐fold cross‐validation. Radiomics models were constructed with the top three frequency features of T1WI/T2WI/two‐sequence MRI (T1WI and FS‐T2WI). Radiomics models, clinical data (age and visually assessed MRI pattern), or radiomics combined with clinical data were used with six classifiers to distinguish between HRC and non‐HRC statuses. Six classifiers used were support vector machine, random forest, logistic regression (LR), decision tree, k‐nearest neighbor, and XGBoost. Model performance was evaluated with area under the curve (AUC) values.Statistical TestsMann–Whitney U‐test, Chi‐squared test, Z test, and DeLong method.ResultsThe LR classifier performed better than the other classifiers based on different data (AUC: 0.65–0.82; P < 0.05). The two‐sequence MRI models performed better than the other data models using different classifiers (AUC: 0.68–0.82; P < 0.05). Thus, the LR two‐sequence model yielded the best performance (AUC: 0.82 ± 0.02; sensitivity: 84.1%; specificity: 68.1%; accuracy: 74.7%; P < 0.05).ConclusionThe LR‐based machine learning method appears superior to other classifier methods for assessing HRC in MM. Radiomics features based on two‐sequence MRI showed good performance in differentiating HRC and non‐HRC statuses in MM.Evidence Level3Technical EfficacyStage 2

Từ khóa


Tài liệu tham khảo

10.1016/S1470-2045(14)70442-5

10.1002/ajh.24402

10.1002/ajh.25791

10.4065/mcp.2009.0603

Kubaczkova V, 2017, Liquid biopsies ‐ the clinics and the molecules, Klin Onkol, 30, 13

10.1016/j.cpet.2019.03.006

10.1186/s12885-018-5024-z

10.1148/radiol.2015151169

10.1007/s00261-019-02028-w

10.7150/thno.30309

10.1007/s00259-020-05049-6

10.1016/j.ejrad.2020.109214

10.3390/cancers12030761

10.1007/s11548-019-02015-y

10.1016/j.ejrad.2019.108739

10.1007/s00330-020-06768-y

10.1002/1097-0142(197509)36:3<842::AID-CNCR2820360303>3.0.CO;2-U

10.1200/JCO.2005.04.242

10.1148/radiology.185.3.1438772

10.1002/ajh.23258

10.1016/j.jcm.2016.02.012

Li W, 2020, Discovery and validation of a CT‐based Radiomic signature for preoperative prediction of early recurrence in Hypopharyngeal carcinoma, Biomed Res Int, 2020

Shtatland ES, 2005, Model building in proc PHREG with automatic variable selection and information criteria, SUGI 30 Proc, 206

10.3390/ijms20163871

10.1111/ele.13538

10.1007/s00330-020-07110-2

10.1038/s41598-020-76827-3

10.1002/jmri.26834

10.1002/jmri.26818

10.21037/atm.2020.01.135

10.1016/j.jacr.2018.09.041

Thông tin
Thông tin xuất bản

Journal of Magnetic Resonance Imaging

Tập 54 Số 4

1303-1311

Thông tin tác giả