Deep Learning Role in Early Diagnosis of Prostate Cancer

被引:67
作者
Reda, Islam [1 ,2 ]
Khalil, Ashraf [3 ]
Elmogy, Mohammed [1 ,2 ]
Abou El-Fetouh, Ahmed [1 ]
Shalaby, Ahmed [2 ]
Abou El-Ghar, Mohamed [4 ]
Elmaghraby, Adel [5 ]
Ghazal, Mohammed [3 ]
El-Baz, Ayman [2 ]
机构
[1] Mansoura Univ, Fac Comp & Informat, Mansoura, Egypt
[2] Univ Louisville, Dept Bioengn, Louisville, KY 40292 USA
[3] Abu Dhabi Univ, Elect & Comp Engn Dept, Abu Dhabi, U Arab Emirates
[4] Mansoura Univ, Radiol Dept, Mansoura, Egypt
[5] Univ Louisville, Dept Comp Engn & Comp Sci, Louisville, KY 40292 USA
关键词
prostate cancer; CAD; PSA; ADC; SNCSAE; APPARENT DIFFUSION-COEFFICIENT; COMPUTER-AIDED DIAGNOSIS; GLEASON SCORE; REPRESENTATION; TISSUE; MRI;
D O I
10.1177/1533034618775530
中图分类号
R73 [肿瘤学];
学科分类号
100214 ;
摘要
The objective of this work is to develop a computer-aided diagnostic system for early diagnosis of prostate cancer. The presented system integrates both clinical biomarkers (prostate-specific antigen) and extracted features from diffusion-weighted magnetic resonance imaging collected at multiple b values. The presented system performs 3 major processing steps. First, prostate delineation using a hybrid approach that combines a level-set model with nonnegative matrix factorization. Second, estimation and normalization of diffusion parameters, which are the apparent diffusion coefficients of the delineated prostate volumes at different b values followed by refinement of those apparent diffusion coefficients using a generalized Gaussian Markov random field model. Then, construction of the cumulative distribution functions of the processed apparent diffusion coefficients at multiple b values. In parallel, a K-nearest neighbor classifier is employed to transform the prostate-specific antigen results into diagnostic probabilities. Finally, those prostate-specific antigen-based probabilities are integrated with the initial diagnostic probabilities obtained using stacked nonnegativity constraint sparse autoencoders that employ apparent diffusion coefficient-cumulative distribution functions for better diagnostic accuracy. Experiments conducted on 18 diffusion-weighted magnetic resonance imaging data sets achieved 94.4% diagnosis accuracy (sensitivity = 88.9% and specificity = 100%), which indicate the promising results of the presented computer-aided diagnostic system.
引用
收藏
页数:11
相关论文
共 32 条
  • [1] [Anonymous], 2009, SIGKDD Explorations, DOI DOI 10.1145/1656274.1656278
  • [2] [Anonymous], 2014, J COMPUTER SCI SYSTE
  • [3] [Anonymous], 2007, P 20 INT C NEURAL IN
  • [4] [Anonymous], ARXIV170304078
  • [5] [Anonymous], SPIE MED IMAGING
  • [6] Applewhite JefireyC., 2000, Cancer control: journal of the Moffitt Cancer Center, V8, P141
  • [7] Bengio Y., ADV NEURAL INFORM PR, P153
  • [8] Representation Learning: A Review and New Perspectives
    Bengio, Yoshua
    Courville, Aaron
    Vincent, Pascal
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2013, 35 (08) : 1798 - 1828
  • [9] Apparent diffusion coefficient ratio correlates significantly with prostate cancer gleason score at final pathology
    Boesen, Lars
    Chabanova, Elizaveta
    Logager, Vibeke
    Balslev, Ingegerd
    Thomsen, Henrik S.
    [J]. JOURNAL OF MAGNETIC RESONANCE IMAGING, 2015, 42 (02) : 446 - 453
  • [10] Neural Decision Forests for Semantic Image Labelling
    Bulo, Samuel Rota
    Kontschieder, Peter
    [J]. 2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2014, : 81 - 88