CoChat
Sign Up
Cited by 1,431Medical Image Segmentation TechniquesAdvanced Neural Network ApplicationsAI in cancer detection

Medical image analysis using convolutional neural networks: a review

Open access
Anwar, Syed Muhammad, Majid, Muhammad, Qayyum, Adnan, Awais, Muhammad, Alnowami, Majdi, Khan, Muhammad Khurram·2018·Journal of Medical Systems
DOI 10.1007/s10916-018-1088-1
Open this paper in CoChat — free accountOpen PDF

Abstract

The science of solving clinical problems by analyzing images generated in clinical practice is known as medical image analysis. The aim is to extract information in an affective and efficient manner for improved clinical diagnosis. The recent advances in the field of biomedical engineering have made medical image analysis one of the top research and development area. One of the reasons for this advancement is the application of machine learning techniques for the analysis of medical images. Deep learning is successfully used as a tool for machine learning, where a neural network is capable of automatically learning features. This is in contrast to those methods where traditionally hand crafted features are used. The selection and calculation of these features is a challenging task. Among deep learning techniques, deep convolutional networks are actively used for the purpose of medical image analysis. This includes application areas such as segmentation, abnormality detection, disease classification, computer aided diagnosis and retrieval. In this study, a comprehensive review of the current state-of-the-art in medical image analysis using deep convolutional networks is presented. The challenges and potential of these techniques are also highlighted.

AI snapshot

Written from the paper’s full text.

TL;DR

This review surveys deep convolutional neural network methods for medical image analysis across segmentation, detection, classification, computer-aided diagnosis, and retrieval. It concludes that CNN-based methods are gaining broad acceptability and often outperform traditional hand-crafted feature approaches, while noting persistent challenges around data, labels, computation, and interpretability.

Key findings

  • CNN-based methods are reported to outperform traditional approaches in cited comparisons. For ILD classification, a CNN achieved precision 92.25, recall 92.21, and F1 92.23, versus HOG + SVM at 76.39, 76.75, and 76.57. For body organ recognition, CNN accuracy was 0.8561 versus random forest 0.7809, K-nearest neighbour 0.7333, and SVM-radial basis function 0.7152.
  • Table 2 reports CNN application accuracies including body part recognition 92.23%, lung texture classification and airway detection 89%, lung pattern classification 85.5%, nuclei detection/classification 80.2%, thyroid nodule diagnosis 83%, breast cancer diagnosis 82.43%, diabetic retinopathy 75%, medical image classification and retrieval 99.77%, radiographic image retrieval 97.79%, and multi-class Alzheimer disease classification 98.88%.
  • For brain tumor segmentation on BRATS, Table 1 reports Dice scores such as 3DNet 3 at 0.91 complete, 0.83 core, and 0.76 enhancing; Cascaded neural networks at 0.86, 0.87, and 0.90; and InputCascade CNN at 0.88, 0.79, and 0.73.
  • The review identifies data scarcity and limited labels as key issues, addressed in the literature through data augmentation and transfer learning; 3D CNN use is described as limited by network size, parameter count, and slow inference.

Methods

  • Design: a comprehensive review of the state-of-the-art in medical image analysis using deep convolutional networks, covering segmentation, abnormality detection, disease classification, computer-aided diagnosis, and retrieval.
  • Data/sample: no primary dataset or new experiments; the paper synthesizes cited studies across imaging modalities including CT, MRI, ultrasound, X-ray, PET, and histology.
  • Analysis: narrative synthesis with summary tables of CNN architectures, clinical applications, performance metrics, and comparisons against other state-of-the-art methods for ILD classification and body organ recognition.

Limitations

  • The paper states that deep learning requires large amounts of training data and computational power; supervised CNNs require labelled data, and manual labelling of medical images is difficult. It also notes the black box problem in AI, noise and illumination problems in medical images, and limited use of 3D CNNs due to network size, parameter count, and slow inference.
  • Evident from the text: as a narrative review, no systematic search strategy, inclusion/exclusion criteria, or quantitative meta-analysis is reported; performance comparisons are drawn from heterogeneous cited studies with varying datasets, numbers of classes, and CNN architectures.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1007/s10916-018-1088-1

Cites

Statistics

Nothing to recompute.

Claims the paper makes

0 of 56 cited sources checked · 0 supported · 0 flagged

  • Findingnone given
    § 4 Medical Image Analysis using CNN · p. 13

    The CNN based method outperforms other methods in major performance indicators for lung pattern classification in ILD disease.

    “The CNN based method outperforms other methods in major performance indicators.”
    Quote in the paperSubstantial claim
  • Findingreasoning
    § 6 Conclusion · p. 15

    It is concluded that convolutional neural network based deep learning methods are finding greater acceptability in all sub-fields of medical image analysis including classification, detection, and segmentation.

    “It is concluded that convolutional neural network based deep learning methods are finding greater acceptability in all sub-fields of medical image analysis including classification, detection, and segmentation.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: In most cases, the data available is limited and expert annotations are scarce.
  • Findingreasoning
    § 5.1 Various Deep Learning Architectures for Medical Image Analysis · p. 14

    DCNN based architectures have found wider success in dealing with medical image data, when compared to other deep learning frameworks.

    “It is generally found that DCNN based architectures have found wider success in dealing with medical image data, when compared to other deep learning frameworks.”
    Quote in the paperNot a substantial claim
  • Findingreasoning
    § 5.1 Various Deep Learning Architectures for Medical Image Analysis · p. 14

    Different DCNN network architectures are adopted or proposed for medical image analysis.

    “This analysis shows that different DCNN network architectures are adopted or proposed for medical image analysis.”
    Quote in the paperSubstantial claim
  • Findingreasoning
    § 4 Medical Image Analysis using CNN · p. 13

    CNN based networks are successful in application areas dealing with multiple modalities for various tasks in medical image analysis.

    “It is seen that CNN based networks are successful in application areas dealing with multiple modalities for various tasks in medical image analysis and provide promising results in almost every case.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim. 45
  • May combine two claims. 26
  • The quote only partly states this claim. 2

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given
    § 6 Conclusion · p. 15

    The problems associated with deep learning techniques due to scarce data and limited labels is addressed by using techniques such as data augmentation and transfer learning.

    “The problems associated with deep learning techniques due to scarce data and limited labels is addressed by using techniques such as data augmentation and transfer learning.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: In most cases, the data available is limited and expert annotations are scarce.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Findingnone given
    § 5.1 Various Deep Learning Architectures for Medical Image Analysis · p. 14

    These architectures focus on reducing the parameter space, improve computation time, and handle 3D data.

    “These architectures focus on reducing the parameter space, improve computation time, and handle 3D data.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Findingnone given
    § 4 Medical Image Analysis using CNN · p. 13

    The results can vary with the number of images used, number of classes, and the choice of the DCNN model.

    “The results can vary with the number of images used, number of classes, and the choice of the DCNN model.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Findingreasoning
    § 4 Medical Image Analysis using CNN · p. 13

    CNN based networks provide promising results in almost every case.

    “It is seen that CNN based networks are successful in application areas dealing with multiple modalities for various tasks in medical image analysis and provide promising results in almost every case.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given
    § 6 Conclusion · p. 15

    For larger datasets, availability of more compute power and better DL architectures is paving the way for a higher performance.

    “For larger datasets, availability of more compute power and better DL architectures is paving the way for a higher performance.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § Medical Image Analysis using Convolutional Neural Networks: A Review · p. 1

    Deep learning is successfully used as a tool for machine learning, where a neural network is capable of automatically learning features.

    “Deep learning is successfully used as a tool for machine learning, where a neural network is capable of automatically learning features.”
    • Also stated in: § 3 Convolutional Neural Networks (CNNs) · p. 7 · plain
      “Deep learning is a tool used for machine learning, where multiple linear as well as non-linear processing units are arranged in a deep architecutre to model high level abstraction present in the data [47].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [47] Deng, L., Yu, D., et al., Deep learning: Methods and applications. Foundations and Trends®, in Signal Processing 7(3–4):197–387, 2014. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 3 Convolutional Neural Networks (CNNs) · p. 7

    In recent years, CNN based methods have gained more popularity in vision systems as well as the medical image analysis domain.

    “In recent years, CNN based methods have gained more popularity in vision systems as well as medical image analysis domain [48,49,50].”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [48] Premaladha, J., and Ravichandran, K., Novel approaches for diagnosing melanoma skin lesions through supervised and deep learning algorithms. J. Med. Syst. 40(4):96, 2016. source

      Not checked in the free run.

    • cites [49] Kharazmi, P., Zheng, J., Lui, H., Wang, Z. J., and Lee, T. K., A computer-aided decision support system for detection and localization of cutaneous vasculature in dermoscopy images via deep feature learning. J. Med. Syst. 42(2):33, 2018. source

      Not checked in the free run.

    • cites [50] Wang, S.-H., Phillips, P., Sui, Y., Liu, B., Yang, M., and Cheng, H., Classification of alzheimer’s disease based on eight-layer convolutional neural network with leaky rectified linear unit and max pooling. J. Med. Syst. 42(5):85, 2018. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A deep CNN for brain tumor segmentation uses a patch based approach with inception method, dropout, batch normalization and inception modules to build the ILinear nexus architecture; dropout regularizer removes over-fitting due to data scarcity.

    “In [58], a deep convolutional neural network is presented for brain tumor segmentation, where a patch based approach with inception method is used for training purpose. Drop-out, batch normalization and inception modules are utilized to build the proposed ILinear nexus architecture. The problem of over-fitting, which arises due to scarcity of data, is removed by using drop-out regularizer.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [58] Hussain, S., Anwar, S. M., and Majid, M., Segmentation of glioma tumors in brain using deep convolutional neural network. Neurocomputing 282:248–261, 2018. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 2.3 Computer Aided Detection or Diagnosis · p. 5

    A hybrid of clinical and image based features is used for multi-class classification of Alzheimer disease using the ADNI dataset with reasonable accuracy.

    “A hybrid of clinical and image based features are used for multi-class classification of alzheimer disease using the alzheimer disease neuro-image initiative (ADNI) dataset with reasonable accuracy [36].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [36] Altaf, T., Anwar, S. M., Gul, N., Majeed, M. N., and Majid, M., Multi-class alzheimer’s disease classification using image and clinical features. Biomed. Signal Process. Control 43:64–74, 2018. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A CNN for brain tumor segmentation uses dropout regularizer to deal with over-fitting and max-out layer as activation function.

    “Another CNN for brain tumor segmentation has been presented in [70]. The architecture uses dropout regularizer to deal with over-fitting, while max-out layer is used as activation function.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [70] Jodoin, A. C., Larochelle, H., Pal, C., and Bengio, Y.: Brain tumor segmentation with deep neural networks

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A CNN for classification of lung disease uses two databases of interstitial lung diseases and CT scans each having dimension 512 × 512, with 14696 image patches derived from original CT scans to train the network.

    “A method for classification of lung disease using a convolutional neural network is presented in [62], which uses two databases of interstitial lung diseases (ILDs) and CT scans each having a dimension of 512 × 512. A total of 14696 image patches are derived from the original CT scans and used to train the network.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [62] Anthimopoulos, M., Christodoulidis, S., Ebner, L., Christe, A., and Mougiakakou, S., Lung pattern classification for interstitial lung diseases using a deep convolutional neural network. IEEE Trans. Med. Imaging 35 (5):1207–1216, 2016. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A two-path eleven-layer deep CNN for brain lesion segmentation is trained using a dense training method with 3D patches, and a 3D fully connected conditional random field is used to remove false positives and perform multiple predictions.

    “A two path eleven layers deep convolutional neural network has been presented in [71] for brain lesion segmentation. The network is trained using a dense training method using 3D patches. A 3D fully connected conditional random field has been used to remove false positives as well as to perform multiple predictions.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [71] Kamnitsas, K., Ledig, C., Newcombe, V. F., Simpson, J. P., Kane, A. D., Menon, D. K., Rueckert, D., and Glocker, B., Efficient multi-scale 3d cnn with fully connected crf for accurate brain lesion segmentation. Med. Image Anal. 36:61–78 , 2017. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given
    § 4 Medical Image Analysis using CNN · p. 9

    Features extracted from techniques such as SIFT are independent of the task or objective function in hand.

    “Features extracted form techniques such as scale invariant feature transform (SIFT) etc. are independent of the task or objective function in hand.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation
    § 2.1 Segmentation · p. 4

    An expectation maximization approach for tumor segmentation on the BRATS 2013 dataset achieves considerable performance but is only tested on a few images and is not shown to generalize for all images in the dataset.

    “An expectation maximization approach is used for tumor segmentation on brain tumor image segmentation (BRATS) 2013 dataset. The method achieves considerable performance, but is only tested on a few images from the dataset and is not shown to generalize for all images in the dataset [25].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [25] Anwar, S., Yousaf, S., and Majid, M.: Brain timor segmentation on multimodal mri scans using emap algorithm. In: Engineering in medicine and biology soceity (EMBC), International Conference of the IEEE. IEEE, pp. 1-4, 2018

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 3 Convolutional Neural Networks (CNNs) · p. 9

    Dropout is used successfully to avoid over-fitting in deep convolutional networks.

    “It has been shown that dropout is used successfully to avoid over-fitting [54].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [54] Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., and Salakhutdinov, R., Dropout: A simple way to prevent neural networks from overfitting. The Journal of Machine Learning Research 15(1):1929–1958, 2014.

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisereasoning
    § 1 Introduction · p. 2

    A major advantage of deep learning methods is their inherent capability to learn complex features directly from raw data, allowing systems that do not rely on hand-crafted features.

    “A major advantage of using deep learning methods is their inherent capability, which allows learning complex features directly from the raw data. This allows us to define a system that does not rely on hand-crafted features, which are mostly required in other machine learning techniques.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 9

    Deep learning algorithms are successfully used for medical image segmentation, computer aided diagnosis, disease detection and classification, and medical image retrieval.

    “Some recent studies have shown that deep learning algorithms are successfully used for medical image segmentation [58], computer aided diagnosis [59,60,61], disease detection and classification [62,63,64,65] and medical image retrieval [66,67].”
    Quote in the paperSubstantial claim0 of 8 sources checked
    • cites [58] Hussain, S., Anwar, S. M., and Majid, M., Segmentation of glioma tumors in brain using deep convolutional neural network. Neurocomputing 282:248–261, 2018. source

      Not checked in the free run.

    • cites [59] Ma, J., Wu, F., Zhu, J., Xu, D., and Kong, D., A pre-trained convolutional neural network based method for thyroid nodule diagnosis. Ultrasonics 73:221–230, 2017. source

      Not checked in the free run.

    • cites [60] Sun, W., Tseng, T.-L. B., Zhang, J., and Qian, W., Enhancing deep convolutional neural network scheme for breast cancer diagnosis with unlabeled data. Comput. Med. Imaging Graph. 57:4–9 , 2017. source

      Not checked in the free run.

    • cites [61] Pratt, H., Coenen, F., Broadbent, D. M., Harding, S. P., and Zheng, Y., Convolutional neural networks for diabetic retinopathy. Procedia Comput. Sci. 90:200–205, 2016. source

      Not checked in the free run.

    • cites [62] Anthimopoulos, M., Christodoulidis, S., Ebner, L., Christe, A., and Mougiakakou, S., Lung pattern classification for interstitial lung diseases using a deep convolutional neural network. IEEE Trans. Med. Imaging 35 (5):1207–1216, 2016. source

      Not checked in the free run.

    • cites [63] van Tulder, G., and de Bruijne, M., Combining generative and discriminative representation learning for lung ct analysis with convolutional restricted boltzmann machines. IEEE Trans. Med. Imaging 35(5):1262–1272, 2016. source

      Not checked in the free run.

    • cites [64] Yan, Z., Zhan, Y., Peng, Z., Liao, S., Shinagawa, Y., Zhang, S., Metaxas, D. N., and Zhou, X. S., Multi-instance deep learning: Discover discriminative local anatomies for bodypart recognition. IEEE Trans. Med. Imaging 35(5):1332–1343, 2016. source

      Not checked in the free run.

    • cites [65] Sirinukunwattana, K., Raza, S. E. A., Tsang, Y.-W., Snead, D. R., Cree, I. A., and Rajpoot, N. M., Locality sensitive deep learning for detection and classification of nuclei in routine colon cancer histology images. IEEE Trans. Med. Imaging 35(5):1196–1206, 2016. source

      Not checked in the free run.

  • May combine two claims.

    Premisereasoning
    § 4 Medical Image Analysis using CNN · p. 9

    Hand-crafted features work when expert knowledge about the field is available and generally make strict assumptions that may not be useful for certain tasks such as medical images.

    “Hand crafted features work when expert knowledge about the field is available and generally make some strict assumptions. These assumptions may not be useful for certain tasks such as medical images.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A convolutional classification restricted Boltzmann machine for lung CT image analysis is trained on 32 × 32 image patches selected along a grid with a 16-voxel overlap; a patch is retained if 75% of voxels belong to the same class.

    “A method based on convolutional classification restricted Boltzmann machine for lung CT image analysis is presented in [63].”
    • Also stated in: § 4 Medical Image Analysis using CNN · p. 11 · plain
      “The network is trained on 32 × 32 image patches selected along a gird with a 16-voxel overlap. A patch is retained if it has 75% of voxel belonging to the same class.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [63] van Tulder, G., and de Bruijne, M., Combining generative and discriminative representation learning for lung ct analysis with convolutional restricted boltzmann machines. IEEE Trans. Med. Imaging 35(5):1262–1272, 2016. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 3 Convolutional Neural Networks (CNNs) · p. 8

    Fractional or overlapping regions for pooling can be beneficial where weak spatial information surrounding the dominant regions of an image is useful.

    “In circumstances where weak spatial information surrounding the dominant regions of an image is also useful, fractional or overlapping regions for pooling could be beneficial [53].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [53] Ding, S., Lin, L., Wang, G., and Chao, H., Deep feature learning with relative distance comparison for person re-identification. Pattern Recog. 48(10):2993–3003, 2015. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given
    § 4 Medical Image Analysis using CNN · p. 9

    In initial layers a DCNN captures edges, blobs and local structure, whereas neurons in higher layers focus on different parts of human organs and some final-layer neurons can consider whole organs.

    “The other advantage is that in the initial layers a DCNN captures edges, blobs and local structure, whereas the neurons in the higher layers focus more on different parts of human organs and some of the neurons in the final layers can consider whole organs.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 3 Convolutional Neural Networks (CNNs) · p. 8

    Pooling layers allow learning features for translational and rotational invariant classification.

    “In addition to down-sampling the feature maps, pooling layers allows learning features for translational and rotational invariant classification [52].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [52] LeCun, Y., Bengio, Y., and Hinton, G., Deep learning. Nature 521(7553):436, 2015. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A CNN based method uses a global-based method taking an entire MRI slice into account to deal with contextual information, and a re-weighting training procedure to deal with data imbalance.

    “The CNN based method presented in [72] deals with the problem of contextual information by using a global-based method, where an entire MRI slice is taken into account in contrast to patch based approach. A re-weighting training procedure has been used to deal with the data imbalance problem.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [72] Tseng, K.-L., Lin, Y.-L., Hsu, W., and Huang, C.-Y.: Joint sequence learning and cross-modality convolution for 3d biomedical segmentation. arXiv: 1704.07754

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A deep learning based approach uses a convolutional layer in place of a fully connected layer to speed up the segmentation process.

    “A deep learning based approach has been presented in [68], in which the network uses a convolutional layer in place of a fully connected layer to speed up the segmentation process.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [68] Havaei, M., Davy, A., Warde-Farley, D., Biard, A., Courville, A., Bengio, Y., Pal, C., Jodoin, P.-M., and Larochelle, H., Brain tumor segmentation with deep neural networks. Med. Image Anal. 35:18–31, 2017. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given
    § 3 Convolutional Neural Networks (CNNs) · p. 9

    The performance of a deep learning method is highly dependent on the data.

    “The performance of a deep learning method is highly dependent on the data.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisereasoning
    § 4 Medical Image Analysis using CNN · p. 9

    A DCNN learns features from the underlying data in an end-to-end learning mechanism, and the error signal from the loss function is propagated back to improve feature extraction, resulting in better representation.

    “On the other hand, a DCNN learn features from the underlying data. These features are data driven and learnt in an end to end learning mechanism. The strength of DCNN is that the error signal obtained by the loss function is used/propagated back to improve the feature (the CNN filters learnt in the initial layers) extraction part and hence, DCNN results in better representation.”
    Quote in the paperNot a substantial claim

  • May combine two claims. The quote only partly states this claim.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    An accuracy of 98.88% is achieved for Alzheimer's disease detection using deep neural networks including GoogLeNet and ResNet on the ADNI dataset, which is higher than traditional machine learning approaches.

    “An accuracy of 98 . 88% is achieved, which is higher than the traditional machine learning approaches used for Alzheimer's disease detection.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [74] Farooq, A., Anwar, S., Awais, M., and Rehman, S.: A deep cnn based multi-class classification of alzheimer’s disease using mri. In: 2017 IEEE International Conference on Imaging systems and techniques (IST). IEEE, pp. 1–6, 2017

      Not checked in the free run.

    • cites [75] Farooq, A., Anwar, S., Awais, M., and Alnowami, M.: Artificial intelligence based smart diagnosis of alzheimer’s disease and mild cognitive impairment. In: 2017 International Smart cities conference (ISC2). IEEE, pp. 1–4, 2017

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A 3D convolutional network for brain tumor segmentation for the BRATS challenge uses a two-path approach to classify each pixel in an MR image.

    “A 3D convolutional network for brain tumor segmentation for the BRATS challenge has been presented in [73]. The network uses a two-path approach to classify each pixel in an MR image.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [73] Casamitjana, A., Puch, S., Aduriz, A., Sayrol, E., and Vilaplana, V.: 3d convolutional networks for brain tumor segmentation. Proceedings of the MICCAI Challenge on Multimodal Brain Tumor Image Segmentation (BRATS), pp. 65–68 , 2016

      Not checked in the free run.

  • May combine two claims.

    Premisecitation
    § 3 Convolutional Neural Networks (CNNs) · p. 9

    Batch normalization speeds up training and in some cases performs regularization, eliminating the need for dropout layers.

    “It is shown that using batch normalization not only speeds up the training but, in some cases, preform regularization eliminating the need for using dropout layers [55].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [55] Ioffe, S., and Szegedy, C.: Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv: 1502.03167

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 10

    A network uses small kernels to classify pixels in MR image; use of small kernels decreases network parameters, allowing deeper networks without worrying about over-fitting.

    “The network presented in [69] uses small kernels to classify pixels in MR image. The use of small kernels decreases network parameters, allowing to build deeper networks, without worrying about the dangers of over-fitting.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [69] Pereira, S., Pinto, A., Alves, V., and Silva, C. A., Brain tumor segmentation using convolutional neural networks in mri images. IEEE Trans. Med. Imaging 35(5):1240–1251 , 2016. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 3 Convolutional Neural Networks (CNNs) · p. 9

    When data availability is limited, various augmentation techniques are utilized.

    “In cases, where the availability of data is limited, various augmentation techniques are utilized [56].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [56] Kooi, T., Litjens, G., van Ginneken, B., Gubern-Mérida, A., Sánchez, C. I., Mann, R., den Heeten, A., and Karssemeijer, N., Large scale deep learning for computer aided detection of mammographic lesions. Med. Image Anal. 35:303–312, 2017. https://doi.org/10.1016/j.media.2016.07.007 . http://www.sciencedirect.com/science/article/pii/S1361841516301244 . source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given
    § 5.3 Limitation of Deep Learning and Future Prospects · p. 15

    One of the main advantages of transfer learning is to enable the use of deeper models to relatively small dataset.

    “One of the main advantages of transfer learning is to enable the use of deeper models to relatively small dataset.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: In most cases, the data available is limited and expert annotations are scarce.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationreasoning
    § 5.2 3D Imaging Modalities · p. 14

    A hybrid of 2D/3D networks and the availability of more compute power is encouraging the use of fully automated 3D network architectures.

    “A hybrid of 2D/3D networks and the availability of more compute power is encouraging the use of fully automated 3D network architectures.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning
    § 6 Conclusion · p. 15

    The recent success indicates that deep learning techniques would greatly benefit the advancement of medical image analysis.

    “The recent success indicates that deep learning techniques would greatly benefit the advancement of medical image analysis.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given
    § 5.3 Limitation of Deep Learning and Future Prospects · p. 15

    Even in the presence of transfer learning more data on the target domain will give better performance.

    “However, even in the presence of transfer learning more data on the target domain will give better performance.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: In most cases, the data available is limited and expert annotations are scarce.

  • May combine two claims.

    Premisecitation
    § 5.2 3D Imaging Modalities · p. 14

    Deep learning methods generally adopt different methods to handle 3D information, including converting 3D volume data into 2D slices and combining features from 2D and multi-view planes.

    “Deep learning methods generally adopt different methods to handle this 3D information. This can involve converting 3D volume data into 2D slices and combination of features from 2D and multi-view planes to benefit from the contextual information [87] [88].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [87] Chen, H., Dou, Q., Yu, L., and Heng, P.-A.: Voxresnet: Deep voxelwise residual networks for volumetric brain segmentation. arXiv: 1608.05895

      Not checked in the free run.

    • cites [88] Setio, A. A. A., Ciompi, F., Litjens, G., Gerke, P., Jacobs, C., Van Riel, S. J., Wille, M. M. W., Naqibullah, M., Sánchez, C. I., and van Ginneken, B., Pulmonary nodule detection in ct images: False positive reduction using multi-view convolutional networks. IEEE Trans. Med. Imaging 35(5):1160–1169, 2016. source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation
    § 1 Introduction · p. 2

    Computer aided detection and computer aided diagnosis rely on effective medical image analysis, making it crucial in terms of performance because it directly affects clinical diagnosis and treatment.

    “The computer aided detection (CADx) and computer aided diagnosis (CAD) relies on effective medical image analysis making it crucial in terms of performance, since it would directly affect the process of clinical diagnosis and treatment [3,4].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [3] Liu, Y., Cheng, H., Huang, J., Zhang, Y., Tang, X., Tian, J.-W., and Wang, Y., Computer aided diagnosis system for breast cancer based on color doppler flow imaging. J. Med. Syst. 36(6):3975–3982, 2012. source

      Not checked in the free run.

    • cites [4] Diao, X.-F., Zhang, X.-Y., Wang, T.-F., Chen, S.-P., Yang, Y., and Zhong, L., Highly sensitive computer aided diagnosis system for breast tumor based on color doppler flow images. J. Med. Syst. 35(5):801–809, 2011. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A framework for body organ recognition based on two-stage multiple instance deep learning extracts discriminative and non-informative patches using CNN in the first stage, then fine-tunes network parameters on extracted discriminative patches in the second stage.

    “In [64], a framework for body organ recognition is presented based on two-stage multiple instance deep learning. In the first stage, discriminative and non-informative patches are extracted using CNN. In the second stage, fine tuning of the network parameters is performed on extracted discriminative patches.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [64] Yan, Z., Zhan, Y., Peng, Z., Liao, S., Shinagawa, Y., Zhang, S., Metaxas, D. N., and Zhou, X. S., Multi-instance deep learning: Discover discriminative local anatomies for bodypart recognition. IEEE Trans. Med. Imaging 35(5):1332–1343, 2016. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given
    § 1 Introduction · p. 2

    Challenges arise due to limited clinical knowledge of DL experts and limited DL knowledge of clinical experts.

    “The challenges arise due to limited clinical knowledge of DL experts and limited DL knowledge of clinical experts.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: There is limited clinical knowledge among deep learning experts and limited deep learning knowledge among clinical experts.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 5.3 Limitation of Deep Learning and Future Prospects · p. 15

    The noise can be removed using pre-processing steps to improve the performance.

    “The noise can be removed using pre-processing steps to improve the performance [58].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [58] Hussain, S., Anwar, S. M., and Majid, M., Segmentation of glioma tumors in brain using deep convolutional neural network. Neurocomputing 282:248–261, 2018. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A deep CNN is proposed to retrieve multimodal images, trained on an intermodal dataset with five modalities and twenty-four classes; three fully connected layers extract features for retrieval.

    “In [66], a deep convolutional neural network has been proposed to retrieve multimodal images. An intermodal dataset having five modalities and twenty-four classes are used to train the network for the purpose of classification. Three fully connected layers are used at the last part of the network for extracting features, which are use for the retrieval.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [66] Qayyum, A., Anwar, S. M., Awais, M., and Majid, M., Medical image retrieval using deep convolutional neural network. Neurocomputing 266:8–20, 2017. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 5.1 Various Deep Learning Architectures for Medical Image Analysis · p. 14

    The success of convolutional neural networks in medical image analysis is evident from a wide spectrum of literature that is recently available.

    “The success of convolutional neural networks in medical image analysis is evident from a wide spectrum of literature that is recently available [79].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [79] Chen, M., Shi, X., Zhang, Y., Wu, D., and Guizani, M.: Deep features learning for medical image analysis with convolutional autoencoder neural network. IEEE Transactions on Big Data (1) 1–1. https://doi.org/10.1109/TBDATA.2017.2717439 , 2017 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A semi-supervised deep CNN based learning scheme for diagnosis of breast cancer is trained on a small set of labeled data.

    “A semi-supervised deep CNN based learning scheme is proposed for the diagnosis of breast cancer[61], and is trained on a small set of labeled data.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [61] Pratt, H., Coenen, F., Broadbent, D. M., Harding, S. P., and Zheng, Y., Convolutional neural networks for diabetic retinopathy. Procedia Comput. Sci. 90:200–205, 2016. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationnone given
    § 5.3 Limitation of Deep Learning and Future Prospects · p. 15

    However, this is partially addressed by using transfer learning.

    “However, this is partially addressed by using transfer learning.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: In most cases, the data available is limited and expert annotations are scarce.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A hybrid thyroid module diagnosis system using two pre-trained CNNs with different numbers of convolutional and fully connected layers uses a soft-max classifier and is validated on 15000 ultrasound images.

    “In [60], a hybrid thyroid module diagnosis system has been proposed by using two pre-trained CNNs. The models differs in terms of the number of convolutional and fully connected layers. A soft-max classifier is used for diagnosis and results are validated on 15000 ultrasound images.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [60] Sun, W., Tseng, T.-L. B., Zhang, J., and Qian, W., Enhancing deep convolutional neural network scheme for breast cancer diagnosis with unlabeled data. Comput. Med. Imaging Graph. 57:4–9 , 2017. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Premisenone given
    § Positron Emission Tomography (PET) · p. 3

    The challenge remains to select an appropriate DL architecture depending upon the number of available images and ground truth labels.

    “The challenge remains to select an appropriate DL architecture depending upon the number of available images and ground truth labels.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The appropriate deep learning architecture depends on the number of available images and ground truth labels.

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationreasoning
    § 1 Introduction · p. 2

    The state-of-the-art in data centric areas such as computer vision shows that deep learning methods could be the most suitable candidate for big data analysis in medical image analysis.

    “As the availability of digital images dealing with clinical information is growing, therefore a method that is best suited to big data analysis is required. The state-of-the-art in data centric areas such as computer vision shows that deep learning methods could be the most suitable candidate for this purpose.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationreasoning
    § 6 Conclusion · p. 15

    This success would ultimately translate into improved computer aided diagnosis and detection systems.

    “This success would ultimately translate into improved computer aided diagnosis and detection systems.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 4 Medical Image Analysis using CNN · p. 11

    A content based medical image retrieval system based on CNN for radiographic images is evaluated using the IRMA database.

    “A content based medical image retrieval (CBMIR) system based on CNN for radiographic images is proposed in [67]. Image retrieval in medical application (IRMA) database is used for the evaluation of the proposed CBMIR system.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [67] Chowdhury, M., Bulo, S. R., Moreno, R., Kundu, M. K., and Smedby, Ö.: An efficient radiographic image retrieval system using convolutional neural network. In: 2016 23rd International Conference on Pattern Recognition (ICPR). IEEE, pp. 3134–3139, 2016

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning
    § 5.3 Limitation of Deep Learning and Future Prospects · p. 15

    Despite the ability of deep learning methods to give better or higher performance, there are some limitations of deep learning techniques, which could limit their application in clinical domain.

    “Despite the ability of deep learning methods to give better or higher performance, there are some limitations of deep learning techniques, which could limit their application in clinical domain.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Deep learning architecture requires a large amount of training data and computational power.
    • Rests on an assumption: Most deep learning techniques such as convolutional neural network requires labelled data for supervised learning and manual labelling of medical images is a difficult task.
    • Rests on an assumption: The application of deep learning in medical image analysis also suffers from the black box problem in AI, where the inputs and outputs are known but the internal representations are not very well understood.
    • Rests on an assumption: These methods are also affected by noise and illumination problems inherent in medical images.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 5.3 Limitation of Deep Learning and Future Prospects · p. 15

    This underlying task for pre-training can be as simple as organ classification or binary classification task of benign or malignant images.

    “This underlying task for pre-training can be as simple as organ classification [66] or binary classification task of benign or malignant images.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [66] Qayyum, A., Anwar, S. M., Awais, M., and Majid, M., Medical image retrieval using deep convolutional neural network. Neurocomputing 266:8–20, 2017. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning
    § 4 Medical Image Analysis using CNN · p. 13

    Convolutional networks will play a crucial role in the development of future medical image analysis systems.

    “Looking at these successes of CNN in medical domain, it seems that convolutional networks will play a crucial role in the development of future medical image analysis systems.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 2 Medical Image Analysis · p. 4

    Machine learning plays a vital role in CADx with applications in tumor segmentation, cancer detection, classification, image guided therapy, medical image annotation, and retrieval.

    “Machine learning plays a vital role in CADx with its applications in tumor segmentation, cancer detection, classification, image guided therapy, medical image annotation, and retrieval [12,13,14,15,16,17,18].”
    Quote in the paperNot a substantial claim0 of 7 sources checked
    • cites [12] Sáez, A., Sánchez-Monedero, J., Gutiérrez, P. A., and Hervás-Martínez, C., Machine learning methods for binary and multiclass classification of melanoma thickness from dermoscopic images. IEEE Trans. Med. Imaging 35(4):1036–1045, 2016. source

      Not checked in the free run.

    • cites [13] Miri, M. S., Abràmoff, M. D., Lee, K., Niemeijer, M., Wang, J.-K., Kwon, Y. H., and Garvin, M. K., Multimodal segmentation of optic disc and cup from sd-oct and color fundus photographs using a machine-learning graph-based approach. IEEE Trans. Med. Imaging 34(9):1854–1866, 2015. source

      Not checked in the free run.

    • cites [14] Gao, Y., Zhan, Y., and Shen, D., Incremental learning with selective memory (ilsm): Towards fast prostate localization for image guided radiotherapy. IEEE Trans. Med. Imaging 33(2):518–534, 2014. source

      Not checked in the free run.

    • cites [15] Tao, Y., Peng, Z., Krishnan, A., and Zhou, X. S., Robust learning-based parsing and annotation of medical radiographs. IEEE Trans. Med. Imaging 30(2):338–350, 2011. source

      Not checked in the free run.

    • cites [16] Ahmad, J., Muhammad, K., Lee, M. Y., and Baik, S. W., Endoscopic image classification and retrieval using clustered convolutional features. J. Med. Syst. 41(12):196, 2017. source

      Not checked in the free run.

    • cites [17] Ahmad, J., Muhammad, K., and Baik, S. W., Medical image retrieval with compact binary codes generated in frequency domain using highly reactive convolutional features. J. Med. Syst. 42(2):24, 2018. source

      Not checked in the free run.

    • cites [18] Jenitta, A., and Ravindran, R. S., Image retrieval based on local mesh vector co-occurrence pattern for medical diagnosis from mri brain images. J. Med. Syst. 41(10):157, 2017. source

      Not checked in the free run.

53 of 53

  • The paper relies on this without support.

    There is limited clinical knowledge among deep learning experts and limited deep learning knowledge among clinical experts.

    “The challenges arise due to limited clinical knowledge of DL experts and limited DL knowledge of clinical experts.”

  • The paper relies on this without support.

    The appropriate deep learning architecture depends on the number of available images and ground truth labels.

    “The challenge remains to select an appropriate DL architecture depending upon the number of available images and ground truth labels.”

  • The paper relies on this without support.

    In most cases, the data available is limited and expert annotations are scarce.

    “In most cases, the data available is limited and expert annotations are scarce.”

  • The paper relies on this without support.

    A large amount of data produced in the medical domain has 3-dimensional information, particularly for volumetric imaging modalities such as CT and MRI.

    “A large amount of data produced in the medical domain has 3-dimensional information. This is particularly true for volumetric imaging modalities such as CT and MRI.”

  • The paper relies on this without support.

    Deep learning architecture requires a large amount of training data and computational power.

    “Deep learning architecture requires a large amount of training data and computational power.”

  • The paper relies on this without support.

    A lack in computational power will lead to a need for more time to train the network, which would depend on the size of training data used.

    “A lack in computational power will lead to a need for more time to train the network, which would depend on the size of training data used.”

  • The paper relies on this without support.

    Most deep learning techniques such as convolutional neural network requires labelled data for supervised learning and manual labelling of medical images is a difficult task.

    “Most deep learning techniques such as convolutional neural network requires labelled data for supervised learning and manual labelling of medical images is a difficult task.”

  • The paper relies on this without support.

    These limitations are being overcome with every passing day due to the availability of more computation power, improved data storage facilities, increasing number of digitally stored medical images and improving architecture of the deep networks.

    “These limitations are being overcome with every passing day due to the availability of more computation power, improved data storage facilities, increasing number of digitally stored medical images and improving architecture of the deep networks.”

  • The paper relies on this without support.

    The application of deep learning in medical image analysis also suffers from the black box problem in AI, where the inputs and outputs are known but the internal representations are not very well understood.

    “The application of deep learning in medical image analysis also suffers from the black box problem in AI, where the inputs and outputs are known but the internal representations are not very well understood.”

  • The paper relies on this without support.

    These methods are also affected by noise and illumination problems inherent in medical images.

    “These methods are also affected by noise and illumination problems inherent in medical images.”

  • The paper relies on this without support.

    The utilization of 3D CNN has been limited in literature due to the size of network and number of parameters involved, leading to slow inference.

    “The utilization of 3D CNN has been limited in literature due to the size of network and number of parameters involved. This also leads to slow inference due to 3D convolutions.”

  • The paper relies on this without support.

    In general, shallow networks are used in situations where data is scarce.

    “In general, shallow networks are used in situations where data is scarce.”

  • The paper relies on this without support.

    In general, a deeper DCNN architecture is the better for the performance.

    “In general, a deeper DCNN architecture is the better for the performance.”
13 of 13

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Traditional statistical performance metrics like the area under the curve may not be sufficient for evaluating models making clinical decisions.

    “Moreover, it has been argued in the literature that traditional statistical performance metrics like the area under the curve may not be sufficient for evaluating the models making clinical decisions [176].”

    From Explainable, trustworthy, and ethical machine learning for healthcare: A survey (2022) · cites this paper as [176]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Thyroid diagnosis is attributed to the breast-cancer reference

    Minor

    The prose assigns the thyroid nodule method to [60], while Table 2 assigns thyroid diagnosis to [59] and breast-cancer diagnosis to [60].

  • Diabetic retinopathy is attributed to the image-retrieval reference

    Minor

    The prose attributes diabetic retinopathy to [66], but Table 2 and the reference list identify [66] as a medical image retrieval study. Table 2 instead assigns diabetic retinopathy to [61].

  • Table 4 is labeled as body-organ recognition but cites lung classification

    Minor

    Table 4 labels its comparison as body-organ recognition, whereas its cited reference [76] is explicitly about classifying lung parenchyma in CT images.

  • Breast-cancer diagnosis is attributed to the diabetic-retinopathy reference

    Minor

    The prose cites [61] for semi-supervised breast-cancer diagnosis, but Table 2 assigns that method to [60] and assigns [61] to diabetic retinopathy.

4 of 4

The citation lists haven’t been fetched yet

References 82 · Cited by 1,431

Research works printed but not linked: 12

  1. Sharma, M. M.: Brain tumor segmentation techniques: A survey. Brain 4 (4): 220–223
  2. Anwar, S., Yousaf, S., and Majid, M.: Brain timor segmentation on multimodal mri scans using emap algorithm. In: Engineering in medicine and biology soceity (EMBC), International Conference of the IEEE. IEEE, pp. 1-4, 2018
  3. Soulami, K. B., Saidi, M. N., and Tamtaoui, A.: A cad system for the detection of abnormalities in the mammograms using the metaheuristic algorithm particle swarm optimization (pso). In: Advances in Ubiquitous Networking 2. Springer, pp. 505–517, 2017
  4. Mizotin, M., Benois-Pineau, J., Allard, M., and Catheline, G.: Feature-based brain mri retrieval for alzheimer disease diagnosis. In: 2012 19th IEEE International Conference on Image Processing (ICIP). IEEE, pp. 1241–1244, 2012
  5. Deng, L., Yu, D., et al., Deep learning: Methods and applications. Foundations and Trends®, in Signal Processing 7(3–4):197–387, 2014. · DOI · Paper page
  6. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., and Salakhutdinov, R., Dropout: A simple way to prevent neural networks from overfitting. The Journal of Machine Learning Research 15(1):1929–1958, 2014.
  7. Casamitjana, A., Puch, S., Aduriz, A., Sayrol, E., and Vilaplana, V.: 3d convolutional networks for brain tumor segmentation. Proceedings of the MICCAI Challenge on Multimodal Brain Tumor Image Segmentation (BRATS), pp. 65–68 , 2016
  8. Gangeh, M. J., Sørensen, L., Shaker, S. B., Kamel, M. S., De Bruijne, M., and Loog, M.: A texton-based approach for the classification of lung parenchyma in ct images. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, pp. 595–602, 2010
  9. Çiçek, Ö., Abdulkadir, A., Lienkamp, S. S., Brox, T., and Ronneberger, O.: 3d u-net: Learning dense volumetric segmentation from sparse annotation. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, pp. 424–432, 2016
  10. Meijs, M., and Manniesing, R.: Artery and vein segmentation of the cerebral vasculature in 4d ct using a 3d fully convolutional neural network. In: Medical Imaging 2018: Computer-Aided Diagnosis, Vol. 10575, International Society for Optics and Photonics, p. 105751Q, 2018
  11. Kamnitsas, K., Ledig, C., Newcombe, V. F., Simpson, J. P., Kane, A. D., Menon, D. K., Rueckert, D., and Glocker, B., Efficient multi-scale 3d cnn with fully connected crf for accurate brain lesion segmentation. Med. Image Anal. 36:61–78, 2017. · DOI · Paper page
  12. Tzeng, E., Hoffman, J., Saenko, K., and Darrell, T.: Adversarial discriminative domain adaptation. In: Computer Vision and Pattern Recognition (CVPR), Vol. 1, p. 4, 2017

Authors
Anwar, S. M., Majid, M., Qayyum, A., Awais, M., Alnowami, M., & Khan, M. K.
Affiliations
  • University of Engineering and Technology Taxila (PK)
  • University of Surrey (GB)
  • King Abdulaziz University (SA)
  • King Saud University (SA)
Corresponding author
Majid, Muhammad
ORCID
Venue
Journal of Medical Systems
Publisher
Springer Science and Business Media LLC
Published
2018-10-08
Volume
42
Issue
11
Pages
226-226
ISSN
  • 0148-5598 (ISSN-L)
  • 1573-689X
DOI
10.1007/s10916-018-1088-1
OpenAlex
W2752517284
Semantic Scholar
4ec06dda674cd5f93351f0ab5871391671f9b974
Cited by
1,431
References
82
Citations by year
2026: 111 · 2025: 222 · 2024: 267 · 2023: 264 · 2022: 201 · 2021: 192 · 2020: 127 · 2019: 36 · 2018: 11
FWCI
10.73
Type
journal article
Language
en
Open access
Licence
www.springer.com/tdm (tdm)
Locations
  • Journal of Medical Systems (publishedVersion)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Medical Image Segmentation Techniques
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
Keywords
medical image analysis, convolutional neural networks, deep learning, image segmentation, anomaly detection, disease classification, computer-aided diagnosis, medical image retrieval, feature extraction
MeSH terms
Deep Learning, Algorithms, Diagnosis, Computer-Assisted, Image Processing, Computer-Assisted, Information Storage and Retrieval, Neural Networks, Computer

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review