2018 · 14 citations · 15 references
ComputingComputer ScienceRadiologyComputer VisionDiagnostic RadiologyMachine LearningComputer-aided DiagnosisDeep LearningLoss FunctionMedical ImagingPancreatic CancerTumor ClassificationClass ImbalanceClass Imbalance LearningComputer-aided Tumor DiagnosisDeep Learning-based Lesion ClassificationDiagnostic EfficiencyLesion ClassificationMedical Image ClassificationPancreatic Tumor Classification
Computer aided diagnosis of pancreatic cancer can help doctors improve diagnostic efficiency and accuracy, which does not depend on the doctor's subjective judgment and experience. Existing pancreatic tumors classification methods suffer from the problem of partial automation and ignore spatial and temporal characteristics. In this paper, we used 3D versions of ResNet18, ResNet34, ResNet52 and Inception-ResNet for pancreatic magnetic resonance images (MRI) classification. In order to alleviate the effect of class imbalance, we proposed a weighted loss function. Two sets of comparative experiments were performed to compare the effect of the proposed loss function. Experimental results show with weighted loss function, the performances of the four models are mostly improved except the precision of ResNet52. Moreover, the false negatives are reduced. Among them, ResNet18 performs best and achieves the accuracy of 91%.
15
Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
Christian Szegedy, Sergey Ioffe, Vincent Vanhoucke et al. · 2017 · 12.6K citations · Full text
Deep Residual Learning for Image Recognition
Kaiming He, Xiangyu Zhang, Shaoqing Ren et al. · arXiv (Cornell University) · 2015 · 4.7K citations · Full text
Holistically-Nested Edge Detection
Saining Xie, Zhuowen Tu · 2015 · 3.1K citations
Holistically-Nested Edge Detection
Saining Xie, Zhuowen Tu · International Journal of Computer Vision · 2017 · 727 citations
Residual and plain convolutional neural networks for 3D brain MRI classification
Sergey Korolev, Amir Safiullin, Mikhail Belyaev et al. · 2017 · 422 citations