Figure 4.
Conventional approaches in comparison with the present method. (a) Comparison of classification accuracy between conventional machine learning algorithms that require hand-crafted features and our 3D deep learning. (b) Performance of deep learning with various input data. SVM: support vector machine; KNN: -nearest neighbors algorithm; LDC: linear discriminant classifier; NB: naïve Bayes; DT: decision tree; MIP: maximum intensity projection.