Discrimination of Malignant and Benign Breast Masses Using Computer-Aided Diagnosis from Dynamic Contrast-Enhanced Magnetic Resonance Imaging

被引:0
|
作者
Ikizceli, Turkan [1 ]
Karacavus, Seyhan [2 ]
Erbay, Hasan [3 ]
Yurttakal, Ahmet Hasim [4 ]
机构
[1] Univ Hlth Sci Turkey, Istanbul Haseki Training & Res Hosp, Clin Radiol, Istanbul, Turkey
[2] Univ Hlth Sci, Dept Nucl Med, Fac Med, Kayseri, Turkey
[3] Turkish Aeronaut Assoc Univ, Dept Comp Engn, Fac Engn, Ankara, Turkey
[4] Afyon Kocatepe Univ, Dept Comp Engn, Fac Engn, Afyon, Turkey
来源
HASEKI TIP BULTENI-MEDICAL BULLETIN OF HASEKI | 2021年 / 59卷 / 03期
关键词
Breast lesions; breast cancer; magnetic resonance imaging; computer-aided diagnosis; segmentation; CANCER; SEGMENTATION; IMAGES; MRI; CLASSIFICATION; ALGORITHM; LESIONS; MODEL;
D O I
10.4274/haseki.galenos.2021.6819
中图分类号
R5 [内科学];
学科分类号
1002 ; 100201 ;
摘要
Aim: To reduce operator dependency and achieve greater accuracy, the computer-aided diagnosis (CAD) systems are becoming a useful tool for detecting noninvasively and determining tissue characterization in medical images. We aimed to suggest a CAD system in discriminating between benign and malignant breast masses. Methods: The dataset was composed of 105 randomly breast magnetic resonance imaging (MRI) including biopsy-proven breast lesions (53 malignant, 52 benign). The expectation-maximization (EM) algorithm was used for image segmentation. 2D-discrete wavelet transform was applied to each region of interests (ROls). After that, intensity-based statistical and texture matrix-based features were extracted from each of the 105 ROIs. Random Forest algorithm was used for feature selection. The final set of features, by random selection base, splatted into two sets as 80% training set (84 MRI) and 20% test set (21 MRI). Three classification algorithms are such that decision tree (DT, C4.5), naive bayes (NB), and linear discriminant analysis (LDA) were used. The accuracy rates of algorithms were compared. Results: C4.5 algorithm classified 20 patients correctly with a success rate of 95.24%. Only one patient was misclassified. The NB classified 19 patients correctly with a success rate of 90.48%. The LDA Algorithm classified 18 patients correctly with a success rate of 85.71%. Conclusion: The CAD equipped with the EM segmentation and C4.5 DT classification was successfully distinguished as benign and malignant breast tumor on MRI.
引用
收藏
页码:190 / 195
页数:6
相关论文
共 50 条
  • [21] Vascularity and Dynamic Contrast-Enhanced Breast Magnetic Resonance Imaging
    Frankhouser, David E.
    Dietze, Eric
    Mahabal, Ashish
    Seewaldt, Victoria L.
    FRONTIERS IN RADIOLOGY, 2021, 1
  • [22] Can Contrast-Enhanced Sonography Distinguish Benign From Malignant Breast Masses?
    Sorelli, P. G.
    Cosgrove, D. O.
    Svensson, W. E.
    Zaman, N.
    Satchithananda, K.
    Barrett, N. K.
    Lim, A. K. P.
    JOURNAL OF CLINICAL ULTRASOUND, 2010, 38 (04) : 177 - 181
  • [23] New Spatiotemporal Features for Improved Discrimination of Benign and Malignant Lesions in Dynamic Contrast-Enhanced-Magnetic Resonance Imaging of the Breast
    Gal, Yaniv
    Mehnert, Andrew
    Bradley, Andrew
    Kennedy, Dominic
    Crozier, Stuart
    JOURNAL OF COMPUTER ASSISTED TOMOGRAPHY, 2011, 35 (05) : 645 - 652
  • [24] Diffusion-weighted imaging or dynamic contrast-enhanced curve: a retrospective analysis of contrast-enhanced magnetic resonance imaging–based differential diagnoses of benign and malignant breast lesions
    Xiaoping Yang
    Mengshi Dong
    Shu Li
    Ruimei Chai
    Zheng Zhang
    Nan Li
    Lina Zhang
    European Radiology, 2020, 30 : 4795 - 4805
  • [25] Semi-Automatic Region-of-Interest Segmentation Based Computer-Aided Diagnosis of Mass Lesions from Dynamic Contrast-Enhanced Magnetic Resonance Imaging Based Breast Cancer Screening
    Levman, Jacob
    Warner, Ellen
    Causer, Petrina
    Martel, Anne
    JOURNAL OF DIGITAL IMAGING, 2014, 27 (05) : 670 - 678
  • [26] The value of dynamic contrast-enhanced magnetic resonance imaging combined with apparent diffusion coefficient in the differentiation of benign and malignant diseases of the breast
    Ao, Feng
    Yan, Yi
    Zhang, Zi-Li
    Li, Sheng
    Li, Wen-Jing
    Chen, Guang-Bin
    ACTA RADIOLOGICA, 2022, 63 (07) : 891 - 900
  • [27] Semi-Automatic Region-of-Interest Segmentation Based Computer-Aided Diagnosis of Mass Lesions from Dynamic Contrast-Enhanced Magnetic Resonance Imaging Based Breast Cancer Screening
    Jacob Levman
    Ellen Warner
    Petrina Causer
    Anne Martel
    Journal of Digital Imaging, 2014, 27 : 670 - 678
  • [28] Histogram analysis of dynamic contrast-enhanced magnetic resonance imaging for differentiating malignant from benign orbital lymphproliferative disorders
    Xu, Xiao-Quan
    Qian, Wen
    Hu, Hao
    Su, Guo-Yi
    Liu, Hu
    Shi, Hai-Bin
    Wu, Fei-Yun
    ACTA RADIOLOGICA, 2019, 60 (02) : 239 - 246
  • [29] Evaluation of quantitative parameters of dynamic contrast-enhanced magnetic resonance imaging in qualitative diagnosis of hepatic masses
    Chen, Jingjing
    Si, Youjiao
    Zhao, Kaikai
    Shi, Xianglong
    Bi, Weiqun
    Liu, Shi-en
    Hua, Hui
    BMC MEDICAL IMAGING, 2018, 18
  • [30] Evaluation of quantitative parameters of dynamic contrast-enhanced magnetic resonance imaging in qualitative diagnosis of hepatic masses
    Jingjing Chen
    Youjiao Si
    Kaikai Zhao
    Xianglong Shi
    Weiqun Bi
    Shi-en Liu
    Hui Hua
    BMC Medical Imaging, 18