2020 · 29 citations · 7 references
ComputingComputer ScienceDermatologyComputer VisionCutaneous OncologyMachine LearningCancer DetectionDeep LearningRepresentation LearningSkin TumorsStaging Of Skin CancerTumor DetectionAccuracyConvolutional Neural Network DetectionDeep Learning DetectionDermoscopic Image AnalysisRegularizationRegularization TechniquesSkin Cancer DetectionSkin Lesion Detection
Melanoma cancer Detection System is a predictive model that dynamically anticipates melanoma skin cancer by evaluating dermoscopic images with the help of deep learning. The fundamental goals behind this research are to identify skin cancer at early stages by achieving swift results With greater accuracy. The reason behind the goal signifies the problem of increment in skin cancer patients Worldwide, high medical costs and exponential increment of death risk for not starting the diagnosis at early stages which is a result of late detection. Our presented research work proposes a solution to the problem of higher medical costs behind diagnosis, lower accuracy rate in detection and portability problem of the manual detection system. In this system, dermoscopic images are classified to predict skin cancer using a multi-layered CNN approach with multiple regularization techniques named dropout and batch normalization. As a result, our system has provided an accuracy of 93.58% which is higher than most other conventional approaches.
7
Dropout: a simple way to prevent neural networks from overfitting
Nitish Srivastava, Geoffrey E. Hinton, Alex Krizhevsky et al. · 2014 · 34.3K citations
Vishal Madan, John T. Lear, Rolf‐Markus Szeimies · The Lancet · 2010 · 876 citations
Medicine, Dermatology, Oncology +16
Does skin cancer screening save lives?
Alexander Katalinic, Annika Waldmann, Martin A. Weinstock et al. · Cancer · 2012 · 277 citations · Full text
Medicine, Dermatology, Oncology +17