The Possibility of Selective Skin Lesion Classification in Convolutional Neural Networks
AbstractSelective classification of skin lesion images and uncertainty estimation is examined to increase the adoption of convolutional neural networks(CNNs) in automated skin cancer diagnostic systems. Research on the application of deep learning models to skin cancer diagnosis has shown success as models outperform medical experts . However, concerns on uncertainty in classifiers and difficulty in approximating uncertainty has caused limited adoption of CNNs in Computer-aided diagnostic systems (CADs) in health care. This research propose selective classification to increase confidence in CNN models for skin cancer diagnosis. The methodology is based on SoftMax response(SR), MC dropout and risk-coverage performance evaluation metric. Risk-coverage curves gives physicians and dermatologist information about the expected rate of misclassification by a model. This enable them to measure the reliability of the classifier’s predictions and inform their decision during skin cancer diagnosis. MC dropout uncertainty estimate was shown to increase accuracy for Melanoma detection by 1.48%. The proposed selective classifier achieved increase melanoma detection. The sensitivity of melanoma increased by 9.91% and 9.73% after selective classification at a coverage of 0.7. This study showed that selective classification and uncertainty estimation can be combined to promote adoption of CNNs in CADs for skin lesions classification.
. Esteva et al, “Dermatologist-level classification of skin cancer with deep nueral networks,” Nature, vol. 542, pp. 115-217, 2017.
. A. Mobiny, A. Singh, and H.V. Nguyen, “Risk-Aware Machine Learning Classifier for skin lesion diagnosis,” Journal of clinical medicine, vol. 8, no. 8, p. 1241, 2019.
. Y. Fujisawa, S. Inuoe, and Y. Nakamura, “The possibility of deep learning based computer-aided skin tumour classifier,” Frontiers in medicine, vol. 6, pp. 1-10, 2019.
. S. Chan et al, “Machine Learning in Dermatology: Current Applications, opportunities and limitations,” Adis Journals, 2020.
. N. Gessert et al, “Skin Lesion Classification Using Loss Balancing and ensemble of Multi-Resolution EfficientNets,” in ISIC, 2019.
. Y. Geifman, and R. El-Yaniv, “Selectivenet: a deep neural network with and integrated reject,” 2019.
. C. Cortes, G. DeSalvo, and M. Mohri, “Boosting with abstention,” In Advances in Neural, pp. 1660-1668, 2016.
. K. Murphy, in Machine learning: A probabilistic perspective, MIT Press, 2012.
. Y. Kwon et al, “Uncertainty quantification using Bayesian neural networks in classification: Application to biomedical image segmentation”.
. Y. Gal, “Uncertainty in deep learning,” University of Cambridge, Cambridge, 2016.
. P.V Molle et al, “Quantifying Uncertainty of Deep Neural Networks in skin lesion classification,” 2019.
. ISIC, “Challenge2019.Isic-Archive,” ISIC. (n.d.), 2019. [Online]. Available: https://challenge2019.isic-archive.com/. [Accessed 2020 05 2020].
. Z. Apalla et al, “Skin cancer: Epidemiology, disease burden, pathophysiology, diagnosis, and therapeutic approches,” Demartol. Ther, pp. 7,5-19, 2017.
. C. R. UK, “Melanoma skin cancer incidence statistics,” 27 06 2020. [Online]. Available: https://www.cancerresearchuk.org/health-professional/cancer-statistics.
. N. C. Institute, “Surveilance, Epidemology and end results program: Cancer stats,” 25 06 2020. [Online]. Available: https://seer.cancer.gov/statfacts.
. I. Goodfellow et al, Deep Learning, MIT Press, 2016.
. E. Hullermeier, and W. Waegeman,, “Aleatoric and Epistemic Uncertainty in Machine,” 2020. [Online]. Available: https://arxiv.org/abs/1910.09457. [Accessed 10 07 2020].
. Y. Gal and Ghahramani, “Bayesian convolutional neural networks with Bernoulli approximate variational inference,” 2016.
. Y. Geifman, and R. El-Yaniv, “Selective classification for deep neural networks,” In Advances in neural information processing systems, no. b, pp. 4878-4887, 2017.
. C. Cortes, G. DeSalvo, and M. Mohri, “Learning with rejection,” In International Conference on Algorithmic Learning Theory, pp. 62-82, 2016.
. K. Hamid, A. Asif, W. Abbasi, D. Sabih, and F. U. A. A. Minhas, “Machine Learning with Abstention for Automated Liver Disease Diagnosis,” International Conference on Frontiers of Information Technology (FIT), Vols. 356-361, 2017.
. M. Tan, and V. Q. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” in ICML, California, 2019.
Copyright (c) 2020 International Journal of Sciences: Basic and Applied Research (IJSBAR)
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Authors who submit papers with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
- By submitting the processing fee, it is understood that the author has agreed to our terms and conditions which may change from time to time without any notice.
- It should be clear for authors that the Editor In Chief is responsible for the final decision about the submitted papers; have the right to accept\reject any paper. The Editor In Chief will choose any option from the following to review the submitted papers:A. send the paper to two reviewers, if the results were negative by one reviewer and positive by the other one; then the editor may send the paper for third reviewer or he take immediately the final decision by accepting\rejecting the paper. The Editor In Chief will ask the selected reviewers to present the results within 7 working days, if they were unable to complete the review within the agreed period then the editor have the right to resend the papers for new reviewers using the same procedure. If the Editor In Chief was not able to find suitable reviewers for certain papers then he have the right to accept\reject the paper.B. sends the paper to a selected editorial board member(s). C. the Editor In Chief himself evaluates the paper.
- Author will take the responsibility what so ever if any copyright infringement or any other violation of any law is done by publishing the research work by the author
- Before publishing, author must check whether this journal is accepted by his employer, or any authority he intends to submit his research work. we will not be responsible in this matter.
- If at any time, due to any legal reason, if the journal stops accepting manuscripts or could not publish already accepted manuscripts, we will have the right to cancel all or any one of the manuscripts without any compensation or returning back any kind of processing cost.
- The cost covered in the publication fees is only for online publication of a single manuscript.