Skip to main content
European Radiology Experimental
Fig. 1 | European Radiology Experimental

Fig. 1

From: Evaluation of techniques to improve a deep learning algorithm for the automatic detection of intracranial haemorrhage on CT head imaging

Fig. 1

Comparison of model performances with different preprocessing and addition of an RNN. CNN denotes the model composed of a CNN, with no preprocessing image techniques or RNN added. CNNwdw denotes the CNN model trained only with the image windowing preprocessing pipeline. CNNslc denotes the CNN model trained only with the adjacent slice concatenation preprocessing pipeline. CNNens denotes the ensemble model combining CNNwdw with CNNslc. CNNwdw-RNN denotes the CNNwdw model joined to an RNN. CNNslc-RNN denotes the CNNslc model joined to an RNN. CNNens-RNN denotes the CNNens model joined to an RNN. AUC-ROC Area under the receiver operating characteristic curve, CNN Convolutional neural network, mAP Average precision score (microaveraged across all six haemorrhage classes), RNN Recurrent neural network

Back to article page