Machine-learning radiomics to predict bone marrow metastasis of neuroblastoma using magnetic resonance imaging

被引:5
|
作者
Lv, Lin [1 ,2 ]
Zhang, Zhengtao [3 ]
Zhang, Dongbo [4 ]
Chen, Qinchang [5 ]
Liu, Yuanfang [6 ]
Qiu, Ya [6 ]
Fu, Wen [3 ]
Yin, Xuntao [7 ]
Chen, Xiong [1 ]
机构
[1] Sun Yat Sen Mem Hosp, Dept Urol Surg, Guangzhou, Guangdong, Peoples R China
[2] Sun Yat Sen Univ, Med Sch, Guangzhou, Guangdong, Peoples R China
[3] Guangzhou Women & Childrens Med Ctr, Guangzhou 510120, Guangdong, Peoples R China
[4] Sun Yat Sen Mem Hosp, Breast Tumor Ctr, Guangzhou, Guangdong, Peoples R China
[5] Guangdong Prov Peoples Hosp, Guangzhou, Guangdong, Peoples R China
[6] Sun Yat Sen Mem Hosp, Dept Radiol, Guangzhou, Guangdong, Peoples R China
[7] Guangzhou Women & Childrens Med Ctr, Dept Radiol, Guangzhou 510120, Guangdong, Peoples R China
来源
CANCER INNOVATION | 2023年 / 2卷 / 05期
基金
中国国家自然科学基金;
关键词
bone marrow metastasis; machine learning; magnetic resonance imaging; neuroblastoma; radiomics; EXTERNAL VALIDATION; MODELS; SYSTEM;
D O I
10.1002/cai2.92
中图分类号
R73 [肿瘤学];
学科分类号
100214 ;
摘要
BackgroundNeuroblastoma is one common pediatric malignancy notorious for high temporal and spatial heterogeneities. More than half of its patients develop distant metastases involving vascularized organs, especially the bone marrow. It is thus necessary to have an economical, noninvasive method without much radiation for follow-ups. Radiomics has been used in many cancers to assist accurate diagnosis but not yet in bone marrow metastasis in neuroblastoma.MethodsA total of 182 patients with neuroblastoma were retrospectively collected and randomly divided into the training and validation sets. Five-hundred and seventy-two radiomics features were extracted from magnetic resonance imaging, among which 41 significant ones were selected via T-test for model development. We attempted 13 machine-learning algorithms and eventually chose three best-performed models. The integrative performance evaluations are based on the area under the curves (AUCs), calibration curves, risk deciles plots, and other indexes.ResultsExtreme gradient boosting, random forest (RF), and adaptive boosting were the top three to predict bone marrow metastases in neuroblastoma while RF was the most accurate one. Its AUC was 0.90 (0.86-0.93), F1 score was 0.82, sensitivity was 0.76, and negative predictive value was 0.79 in the training set. The values were 0.82 (0.71-0.93), 0.80, 0.75, and 0.92 in the validation set, respectively.ConclusionsRadiomics models are likely to contribute more to metastatic diagnoses and the formulation of personalized healthcare strategies in clinics. It has great potential of being a revolutionary method to replace traditional interventions in the future. A total of 182 patients with neuroblastoma were retrospectively collected and randomly divided into the training and validation sets. Radiomics features were extracted from magnetic resonance imaging, significant features were selected via T-test for model development. We attempted 13 machine-learning algorithms and eventually chose three best-performed models. The integrative performance evaluations are based on the area under the curve, calibration curves, risk deciles plots, and other indexes. The top three performance models were extreme gradient boosting, random forest, and adaptive boosting. image
引用
收藏
页码:405 / 415
页数:11
相关论文
共 50 条
  • [31] MAGNETIC-RESONANCE-IMAGING OF BONE-MARROW
    PATHRIA, MN
    ISSACS, P
    CURRENT OPINION IN RADIOLOGY, 1992, 4 (06) : 21 - 31
  • [32] Magnetic resonance imaging of the bone marrow in hematologic malignancies
    Moulopoulos, LA
    Dimopoulos, MA
    BLOOD, 1997, 90 (06) : 2127 - 2147
  • [33] Magnetic resonance imaging of bone marrow for TAFRO syndrome
    Nakamura, Gen
    Homma, Noriyuki
    Kasai, Akio
    Kasami, Takuya
    Makino, Kunihiko
    Aoki, Youhei
    Wakaki, Kunihiko
    Nakagawa, Norihito
    MODERN RHEUMATOLOGY, 2019, 29 (03) : 551 - 557
  • [34] Magnetic resonance imaging of the bone marrow in hematological malignancies
    Vande Berg, BC
    Lecouvet, FE
    Michaux, L
    Ferrant, A
    Maldague, B
    Malghem, J
    EUROPEAN RADIOLOGY, 1998, 8 (08) : 1335 - +
  • [35] Evaluation of bone marrow iron by magnetic resonance imaging
    Isokawa, M
    Kimura, F
    Matsuki, T
    Omoto, E
    Otsuka, K
    Kurokawa, H
    Togami, I
    Hiraki, Y
    Kimura, I
    Harada, M
    ANNALS OF HEMATOLOGY, 1997, 74 (06) : 269 - 274
  • [36] Magnetic resonance imaging of the bone marrow in hematological malignancies
    B. C. Vande Berg
    F. E. Lecouvet
    L. Michaux
    A. Ferrant
    B. Maldague
    J. Malghem
    European Radiology, 1998, 8 : 1335 - 1344
  • [37] Magnetic resonance imaging of bone marrow in hematologic malignancies
    Takagi, S
    INTERNATIONAL JOURNAL OF HEMATOLOGY, 1997, 66 (04) : 413 - 422
  • [38] Evaluation of bone marrow iron by magnetic resonance imaging
    M. Isokawa
    F. Kimura
    T. Matsuki
    E. Omoto
    K. Otsuka
    H. Kurokawa
    I. Togami
    Y. Hiraki
    I. Kimura
    M. Harada
    Annals of Hematology, 1997, 74 : 269 - 274
  • [39] Magnetic resonance imaging of the bone marrow in lymphomas and leukemias
    Tardivon, AA
    Vanel, D
    Munck, JN
    Bosq, J
    LEUKEMIA & LYMPHOMA, 1997, 25 (1-2) : 55 - 68
  • [40] Comparison of two methods for evaluating bone marrow metastasis of neuroblastoma: Reverse transcription-polymerase chain reaction for tyrosine hydroxylase and magnetic resonance imaging
    Takemoto, C
    Nishiuchi, R
    Endo, C
    Oda, M
    Seino, Y
    PEDIATRICS INTERNATIONAL, 2004, 46 (04) : 387 - 393