Intracranial hemorrhage subtype classification using learned fully connected separable convolutional network

Abstract

Summary In recent decades, intracranial hemorrhage detection from computed tomography (CT) scans has gained considerable attention among researchers in the medical community. The major problem in dealing with the Radiological Society of North America (RSNA) dataset is a three dimensional representation of CT scan, where the labeled data are scarce and hard to obtain. To highlight this problem, a novel learned fully connected separable convolutional network is proposed in this research article. After collecting the CT scans, data augmentation is used to generate multiple image variations to improve the capacity of the proposed model generalization. Based on the albumentations library, the transformations are selected for data augmentation such as brightness adjustment, horizontal flipping, shifting, rotation, and scaling. The intracranial hemorrhage subtype classification is accomplished utilizing a learned fully connected separable convolutional network which significantly classifies six classes as any, intraparenchymal, subarachnoid, epidural, intraventricular, and subdural. In the resulting phase, the learned fully connected separable convolutional network obtained an average accuracy of 98.63%, sensitivity of 73.32%, specificity of 99.49%, and area under the curve of 98.98%, where the obtained results are effective compared with ResNet‐50, SE‐ResNeXt‐50, ResNeXt‐101, and ResNeXt‐101 with bidirectional long short term memory network.

Performance Evaluation of Different Deep Learning Models for Classifying Ischemic, Hemorrhagic, and Normal Computed Tomography Images: Transfer Learning Approaches

“…The average values of the area under the curve were calculated as 0.9920±0.005 for ischemia, 0.9828±0.008 for hemorrhage, and 0.9686±0.012 for normal using AlexNet. There are studies in the literature for the detection and classification of stroke with deep learning models and CT images [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28]. However, many studies classify either hemorrhage or ischemia.…”

“…In the literature, some studies classify stroke using deep learning algorithms on CT images. However, many of these studies are focused on classifying either ischemic or hemorrhagic strokes [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28]. Fewer studies simultaneously classify normal, ischemic, and hemorrhagic stroke images [9,12,14,23].…”

A real-time automated bone age assessment system based on the RUS-CHN method

“…The second part of the system chose to use Lightgbm to calculate bone age with an average elapsed time of 2 ms. Finally, the first part of the system was selected as YOLOv5, the confidence threshold of KBS was selected as 0.1 and pre-processing was performed with the help of the Albumentations ( 36 ). The average processing time in the environment of GPU RTX3060 is 26 ms.…”

Automated Computer-Aided Detection and Classification of Intracranial Hemorrhage Using Ensemble Deep Learning Techniques

“…There are 4096 values (12 bits), and the scale ranges from −1024 HU to 3071 HU (zero is also a value). Using this scale, it is possible to correlate the attenuation of CTs with the density of tissues [ 28 ]. Using the “windowing” technique, CT images are enhanced based on their contrast.…”