The value of machine learning in detection and assessment of liver trauma using contrast enhanced CT scans

Sweed, Enas; El Sheikh, Hesham; Al-Tohamy, Jehan; Sharaf Eldeen, Marwa Khaled; Ahmed, Khaled

doi:10.21608/bjas.2025.349604.1564

The value of machine learning in detection and assessment of liver trauma using contrast enhanced CT scans

Document Type : Original Research Papers

Authors

¹ Radiodiagnosis Department, Faculty of Medicine, Benha University, Benha, Egypt.

² Professor of Diagnostic Radiology, Faculty of Medicine, Benha University, Egypt

³ Lecturer of Radiology Armed forces college of medicine AFCM and General organization for teaching hospitals and institutes GOTHI, Egypt

⁴ MScs, assistant lecturer of Diagnostic Radiology, Faculty of Medicine, Benha University

⁵ Lecturer of Bio-medical Engineering, Faculty of Engineering, Benha University, Egypt

10.21608/bjas.2025.349604.1564

Abstract

Background: Over the last few years, there has been increasing interest in the use of artificial intelligence to assist with abnormality detection on medical images. Aim of this study was to examine how well AI performed when used with High Resolution Multi-Slice Computed Tomography to detect liver trauma. Methods: This prospective study examined 65 instances that were analysed using computer-aided detection techniques to automatically identify liver trauma. The study's main emphasis was the application of artificial intelligence technologies in liver multislice CT images. To check for abnormalities in the liver, we used High Resolution Multi slice Computed Tomography, which has a 16/64/128 detector and is powered by artificial intelligence. Findings: a median age of 20 and a range of 15 years for the standard deviation. Artificial intelligence's diagnostic ability in identifying liver lesions was examined. When it came to identifying liver lesions, the models were in high agreement. Conclusion: The liver and traumatised areas can be precisely segmented using the techniques that have been suggested. It is clear that this pipeline performs admirably when it comes to determining the proportion of traumatised liver parenchyma. Rather of relying on the sometimes-subjective AAST grading system, critical care medical staff can benefit from this approach's reproducible quantitative evaluation of liver injuries.

Keywords

Main Subjects