Classification of Intracranial Hemorrhage Subtypes Using Deep Learning on CT Scans

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Neurosurgery

Background:

• Intracranial hemorrhage necessitates prompt diagnosis and treatment.
• Machine learning models offer potential for automating medical image analysis.
• A ResNeXt-based neural network was previously developed for hemorrhage classification.

Purpose of the Study:

• To evaluate the performance of a deep learning model for intracranial hemorrhage classification on real-world CT scans.
• To assess the model's accuracy in detecting different subtypes of intracranial hemorrhage.

Main Methods:

• Utilized a pre-trained deep learning model based on ResNeXt architecture.
• Tested the model on a dataset of CT scans from patients with intracranial hemorrhage.
• Evaluated the model's detection accuracy for each hemorrhage subtype without parameter tuning.

Main Results:

• The deep learning model achieved an accuracy greater than 0.81 for all intracranial hemorrhage subtypes.
• The model demonstrated robust performance on real-world clinical data.

Conclusions:

• The ResNeXt-based deep learning model shows significant potential for accurate and rapid classification of intracranial hemorrhage.
• Further improvements in model performance are anticipated with additional tuning and larger datasets.