Improving Sensitivity on Identification and Delineation of Intracranial Hemorrhage Lesion Using Cascaded Deep Learning Models

Abstract

Highly accurate detection of the intracranial hemorrhage without delay is a critical clinical issue for the diagnostic decision and treatment in an emergency room. In the context of a study on diagnostic accuracy, there is a tradeoff between sensitivity and specificity. In order to improve sensitivity while preserving specificity, we propose a cascade deep learning model constructed using two convolutional neural networks (CNNs) and dual fully convolutional networks (FCNs). The cascade CNN model is built for identifying bleeding; hereafter the dual FCN is to detect five different subtypes of intracranial hemorrhage and to delineate their lesions. Using a total of 135,974 CT images including 33,391 images labeled as bleeding, each of CNN/FCN models was trained separately on image data preprocessed by two different settings of window level/width. One is a default window (50/100[level/width]) and the other is a stroke window setting (40/40). By combining them, we obtained a better outcome on both binary classification and segmentation of hemorrhagic lesions compared to a single CNN and FCN model. In determining whether it is bleeding or not, there was around 1% improvement in sensitivity (97.91% [± 0.47]) while retaining specificity (98.76% [± 0.10]). For delineation of bleeding lesions, we obtained overall segmentation performance at 80.19% in precision and 82.15% in recall which is 3.44% improvement compared to using a single FCN model.

Intracranial Hemorrhage Segmentation Using a Deep Convolutional Model

“…The other deep learning-based models in Ref. [15,17,19,23] were trained and tested on larger datasets and achieved higher performance for the ICH segmentation. Ref.…”

“…[17] reported a 78% overlap between the attention maps of their CNN model and the gold-standard bleeding points, Ref. [23] reported 78% average precision, and [19] reported 80.19% precision and 82.15% recall. In addition to the deep learning methods, in the study of Ref.…”

“…Some researchers have extended the scope and performed the ICH segmentation to identify the region of ICH [7,11,15,17,[19][20][21][22][23][24][25][26]. Most researchers validated their algorithms using small datasets [7][8][9][10][11][12][13]17,[20][21][22][24][25][26], while a few used large datasets for testing and validating [6,[14][15][16]18,19,23]. We provide a comprehensive review of the published papers for the ICH detection and segmentation ( Figure 1) in this section.…”

“…Some authors also modified CNN to process some part or the entire CT scan [14,15] or used an interpolation layer [17]. Other approaches were 1-stage , meaning that they did not consider the spatial dependency between the slices [12,19,23]. Prevedello and colleagues proposed two algorithms based on CNN [12].…”

“…The application of the cost-sensitive active learning technique improved the model performance on the prospective test set by annotating the new CT scans and increasing the size of the training data/scans. In [19], the CNN cascade model was used for the ICH detection and the dual FCN models was used for the ICH segmentation. The CNN cascade model was based on the GoogLeNet network, and the dual FCN model was based on a pre-trained VGG16 network that was modified and fine-tuned on the brain and stroke window settings.…”