Fig. 3From: Machine learning combined with radiomics and deep learning features extracted from CT images: a novel AI model to distinguish benign from malignant ovarian tumorsThe architecture of the 3D U-net used for DL feature extraction. The architecture includes an encoder network and a decoder network. The encoder extracts tumor characteristics referred to as DL features, and the decoder uses the DL features to reconstruct original tumor image. The segmented tumor images were input into the network. The output of the last convolutional layer in the encoder network was extracted as a 224-dimensional DL featureBack to article page