Abstract
Convolutional neural networks (CNNs) have been broadly employed in dermoscopic image analysis, mainly as a result of the large amount of data gathered by the International Skin Imaging Collaboration (ISIC). As in many other medical imaging domains, state-of-the-art methods take advantage of architectures developed for other tasks, frequently assuming full transferability between enormous sets of natural images (e.g. ImageNet) and dermoscopic images, which is not always the case. A comprehensive analysis on the effectiveness of state-of-the-art deep learning techniques when applied to dermoscopic image analysis is provided. To achieve this goal, the authors consider several CNNs architectures and analyse how their performance is affected by the size of the network, image resolution, data augmentation process, amount of available data, and model calibration. Moreover, taking advantage of the analysis performed, a novel ensemble method to further increase the classification accuracy is designed. The proposed solution achieved the third best result in the 2019 official ISIC challenge, with an accuracy of 0.593.
Highlights
Skin cancer is a major public health issue, being the most common forms of human cancer worldwide [1]
- We perform a thorough investigation about the performance of state-of-the-art architectures for natural images classification when applied to dermoscopic image analysis. - A comprehensive discussion on how the major hyperparameters affect neural network capabilities is provided. - We explore and motivate the use of model calibration in order to improve the overall accuracy of a deep learning architecture in skin lesion analysis. - We design a novel ensemble method for dermoscopic image classification, which yields a balanced accuracy of 0.593 on the official 2019 International Skin Imaging Collaboration (ISIC) challenge, achieving the third best result. - The first classified algorithm of the official 2019 ISIC challenge is compared with the proposed method
With this paper we addressed the impact of image resolution, data augmentation, and different state-of-the-art architectures on dermoscopic images analysis
Summary
Skin cancer is a major public health issue, being the most common forms of human cancer worldwide [1]. To perform a fast diagnosis, many dermatologists rely on dermoscopy, which is a form of in-vivo skin surface microscopy performed using high quality magnifying lenses and a powerful light source to mitigate the surface reflection of the skin, in order to enhance the visibility of the pigmentation of the lesion (Fig. 1 and Fig. 2) This imaging technique has increased the diagnosis accuracy, sensitivity, and specificity with respect to the naked eye examination, mitigating the need of surgical intervention for the unnecessary removal of benign lesions. To diagnose skin cancer through this kind of non-invasive imaging approaches, a thorough image analysis must be performed by expert clinicians. This is why many efforts have been given in recent years towards the creation of tools to assist physicians in the analysis of dermoscopic images. Due to its outstanding results in many areas such as speech recognition [3], image understanding [4] and image classification [5, 6], has become the main option for analyzing medical images
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.
