Abstract #2522

Detection of liver fibrosis from MRIusing histogram of strains

Yasmine A. Safwat¹, Rasha S. Hussein², Ayman Khalifa³, Ahmed S. Ibrahim⁴, Ahmed Samir⁵, Heba Abdallah⁶, and Ahmed S Fahmy⁷

¹Center for Informatics Science, Nile University, Cairo, Egypt, ²Radiodiagnosis, Ain Shams University, Cairo, Egypt, ³Biomedical Engineering Department, Helwan University, Cairo, Egypt, ⁴Radiodiagnosi, Ain Shams University, Cairo, Egypt, ⁵Tropical Department, Ain Shams University, Cairo, Egypt, ⁶Tropical Medicine Department, Ain Shams University, Cairo, Egypt, ⁷Biomedical Engineering Department, Cairo University, Cairo, Egypt

In this work, we present the results of a novel method for detecting liver fibrosis from tagged MRI images. The method is based on extracting a set of features representing the liver deformations induced by the heart motion. First, the tagged MRI images are analyzed to calculate the liver tissue strain induced by the heart motion. The histogram of the peak strain values at each point within the liver are used as feature vectors to classify normal from patients with liver fibrosis. Classification using support-vector-machines using data of 34 subject (15 normal, 19 patients) showed sensitivity and specificity of 89%, and 80% respectively.

This abstract and the presentation materials are available to members only; a login is required.

Join Here