Related Experiment Video
Updated: Jan 1, 2026
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
43.4K
[Feasibility Study on Location of CT Images Using Convolutional Neural Networks]
Hang Yu1, Jialiang Jiang1, Yisong He1
1Department of Radiotherapy, West China Hospital of Sichuan University, Chengdu, 610041.
Summary
Deep learning using convolutional neural networks can accurately locate CT images. Data augmentation enhances classification accuracy but slightly increases processing time for medical imaging analysis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Accurate localization of anatomical structures in CT images is crucial for diagnosis.
- Deep learning models offer potential for automating image analysis tasks.
Purpose of the Study:
- To evaluate the efficacy of a deep learning model, specifically AlexNet, for locating CT images.
- To assess the impact of data augmentation on classification accuracy and processing time.
Main Methods:
- Utilized the AlexNet convolutional neural network architecture.
- Employed transfer learning for model pre-training.
- Categorized and labeled training samples based on vertebral body parts.
- Applied data augmentation techniques to enhance model performance.
Main Results:
- Achieved a classification accuracy of 97.72% after data augmentation, an improvement from 94.95%.
- Observed an increase in testing time from 2.05 seconds to 3.03 seconds due to augmentation.
Conclusions:
- Convolutional neural networks are a feasible tool for CT image localization.
- Data augmentation effectively improves classification accuracy in CT image analysis.
- The benefits of increased accuracy must be weighed against the rise in computational time.
