Document Type

Conference Item

Publication Date

1-1-2021

Abstract

Application of texture-based radiomics to breast magnetic resonance images (MRI) for predicting and diagnosing breast cancer.84 lesions with histopathologically confirmed primary breast cancer were retrospectively evaluated. 3D volumetric breast lesions were segmented from subtracted dynamic contrast-enhanced (DCE) images. The segmented 3D mask (region of interest) was applied to different MRI sequences (T1-weighted, T2-weighted, STIR, DCE Phase 2, subtracted Phase 2) for texture analysis (TA) of the lesion using MATLAB software. TA of contralateral normal breast tissues were also performed for comparison. The texture features were used to analyze and classify immunohistochemical subtypes, histopathological grades and MRI kinetic curves using Kruskal-Wallis test and Random Forest classification. Validation of radiomics model was carried out using leave one-out-crossvalidation.

Keywords

Magnetic resonance imaging, Texture analysis, Breast cancer

Divisions

biomed

Event Title

19th Asian Oceanian Congress of Radiology Incorporating Malaysian Congress of Radiology

Event Location

Kuala Lumpur (Virtual)

Event Dates

1-4 July 2021

Event Type

conference

Share

COinS