IBM Journal of Research and Development · 2017 · 86 citations · 27 references
ComputingComputer ScienceDermatologyComputer VisionDiagnostic ImagingMachine LearningCancer DetectionDeep LearningEnsemble LearningEnsemble MethodsSkin TumorsTumor DetectionArea Under The Receiver Operating Characteristic CurveBenchmark Dataset DevelopmentBenchmark Dataset EvaluationDeep EnsemblesDeep Learning DetectionDermoscopic Image AnalysisMelanoma DiagnosisSkin Cancer Detection
Melanoma is the deadliest skin cancer, and early detection by specialists is essential for cure. Automated melanoma detection could save lives, reduce unnecessary biopsies, and cut costs. We combine deep‑learning segmentation with machine‑learning analysis in ensembles, evaluated on the largest dermoscopy benchmark of 900 training and 379 test images. Our system achieves state‑of‑the‑art performance, raising AUROC by 7.5 % to 0.843, average precision by 4 % to 0.649, and specificity at 95 % sensitivity 2.9× higher (36.8 % vs 12.5 %), outperforming eight dermatologists on 100 test images (76 % vs 70.5 % accuracy, 62 % vs 59 % specificity).
Melanoma is the deadliest form of skin cancer. While curable with early detection, only highly trained specialists are capable of accurately recognizing the disease. As expertise is in limited supply, automated systems capable of identifying disease could save lives, reduce unnecessary biopsies, and reduce costs. Toward this goal, we propose a system that combines recent developments in deep learning with established machine learning approaches, creating ensembles of methods that are capable of segmenting skin lesions, as well as analyzing the detected area and surrounding tissue for melanoma detection. The system is evaluated using the largest publicly available benchmark dataset of dermoscopic images, containing 900 training and 379 testing images. New state-of-the-art performance levels are demonstrated, leading to an improvement in the area under receiver operating characteristic curve of 7.5% (0.843 versus 0.783), in average precision of 4% (0.649 versus 0.624), and in specificity measured at the clinically relevant 95% sensitivity operating point 2.9 times higher than the previous state of the art (36.8% specificity compared to 12.5%). Compared to the average of eight expert dermatologists on a subset of 100 test images, the proposed system produces a higher accuracy (76% versus 70.5%), and specificity (62% versus 59%) evaluated at an equivalent sensitivity (82%).
27
Deep Residual Learning for Image Recognition
Kaiming He, Xiangyu Zhang, Shaoqing Ren et al. · 2016 · 222.1K citations · Full text
Fully convolutional networks for semantic segmentation
Jonathan Long, Evan Shelhamer, Trevor Darrell · 2015 · 36.9K citations
Yangqing Jia, Evan Shelhamer, Jeff Donahue et al. · 2014 · 11.2K citations
FaceNet: A unified embedding for face recognition and clustering
2015 · 11.2K citations · Full text
DeepFace: Closing the Gap to Human-Level Performance in Face Verification
Yaniv Taigman, Ming Yang, Marc’Aurelio Ranzato et al. · 2014 · 6.6K citations