Abstract
Melanoma is one of the main causes of cancer-related deaths. The development of new computational methods as an important tool for assisting doctors can lead to early diagnosis and effectively reduce mortality. In this work, we propose a convolutional neural network architecture for melanoma diagnosis inspired by ensemble learning and genetic algorithms. The architecture is designed by a genetic algorithm that finds optimal members of the ensemble. Additionally, the abstract features of all models are merged and, as a result, additional prediction capabilities are obtained. The diagnosis is achieved by combining all individual predictions. In this manner, the training process is implicitly regularized, showing better convergence, mitigating the overfitting of the model, and improving the generalization performance. The aim is to find the models that best contribute to the ensemble. The proposed approach also leverages data augmentation, transfer learning, and a segmentation algorithm. The segmentation can be performed without training and with a central processing unit, thus avoiding a significant amount of computational power, while maintaining its competitive performance. To evaluate the proposal, an extensive experimental study was conducted on sixteen skin image datasets, where state-of-the-art models were significantly outperformed. This study corroborated that genetic algorithms can be employed to effectively find suitable architectures for the diagnosis of melanoma, achieving in overall 11% and 13% better prediction performances compared to the closest model in dermoscopic and non-dermoscopic images, respectively. Finally, the proposal was implemented in a web application in order to assist dermatologists and it can be consulted at http://skinensemble.com.
Highlights
Melanoma is the most serious form of skin cancer that begins in cells known as melanocytes
The results showed that the proposed approach achieved promising results and was competitive compared to six state-of-the-art convolutional neural networks (CNNs) models which have previously been used for diagnosing melanoma [3, 10, 24,25,26]
In PH2 no significant differences were encountered between the three methods, the Friedman’s test did not reject the null hypothesis with a p-value equal to 6.476E-1. (The test was conducted with two degrees of freedom, and the Friedman’s statistic was equal to 86.898E-2.) the proposal was ranked first and it attained 111% and 74% less variance compared to U-Net and R2U-Net, respectively
Summary
Melanoma is the most serious form of skin cancer that begins in cells known as melanocytes. Melanoma has an increasing incidence, where just in Europe were estimated 144,200 cases and 20,000 deaths in 2018 [1], whereas in USA, 106,110 new cases of invasive melanoma will be Several automated computer image analysis strategies have been used as tools for medical practitioners to provide accurate lesion diagnostics, including descriptor-based methods [5, 6] and convolutional neural networks (CNNs) [3, 7, 8]. Neural Computing and Applications extraction of handcrafted features [9], which rely on the expertise of dermatologists and introduce a margin of error. CNN models can automatically learn highlevel features from raw images [3], allowing for the development of applications in a shorter timeframe. Nasr-Esfahani et al [10] showed that CNN models can overcome handcrafted features-based methods. Brinker et al [11] demonstrated that CNN models can match the prediction performance of 145 dermatologists
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
