Related Experiment Video
Updated: Jan 17, 2026
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
43.4K
Integrating CT image reconstruction, segmentation, and large language models for enhanced diagnostic insight
Altamash Ahmad Abbasi1, Ashfaq Hussain Farooqi2
1Department of Computer Science, Air University, Islamabad, Pakistan.
Medical & Biological Engineering & Computing
|September 24, 2025
Summary
This study introduces a deep learning framework to enhance computed tomography (CT) image quality and speed up reconstruction. The system improves diagnostic accuracy by providing clearer medical images and automated descriptions for faster clinical decisions.
Area of Science:
- Medical Imaging
- Deep Learning Applications
- Radiology
Background:
- Computed Tomography (CT) is crucial for diagnosing and monitoring patients with heart and cancer conditions.
- High-quality CT images are essential for accurate clinical decision-making.
- Image reconstruction is a key research area for improving CT image quality.
Purpose of the Study:
- To develop a framework for enhancing CT image quality.
- To minimize CT image reconstruction time.
- To create a decision-support tool for medical experts.
Main Methods:
- A four-step framework: reconstruction, preprocessing, segmentation, and image description.
- Utilized Radon transform for sinogram generation and Convolutional Neural Network (CNN) for high-quality reconstruction.
- Applied bilateral filtering for noise reduction, K-means clustering for segmentation, and FuseCap model for automated image description.
Main Results:
- Achieved high performance metrics: Peak Signal-to-Noise Ratio (PSNR) of 30.784, Normalized Mean Square Error (NMSE) of 0.032, and Structural Similarity Index Measure (SSIM) of 0.877.
- Demonstrated superior performance compared to existing CT image reconstruction methods.
- The framework successfully reconstructed high-quality CT images with integrated segmentation and automated descriptions.
Conclusions:
- The proposed framework effectively reconstructs high-quality CT images from raw projection data.
- The integration of segmentation and automated descriptions enhances diagnostic efficiency.
- This research aims to reduce medical professionals' workload and improve patient care through advanced imaging analysis.
