Publication: Classification of glioma brain tumors in mr images using imagej-based radiomic analysis
No Thumbnail Available
Date
2024-07
Authors
Zafrin, Nurizzatul Hadawiyah Mohamed
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The aim of this study is to classify glioma brain tumor grades using ImageJ-based radiomic analysis. This study utilizes magnetic resonance imaging (MRI) with ImageJ (Fiji) software to perform radiomic analysis, providing a quantitative method to evaluate Glioma tumor without the use of Gadolinium based contrast agents (GBCAs). Gliomas can range from low grade gliomas (LGGs) (Grades I and II) to high grade gliomas (HGGs) (Grades III and IV) tumors. Accurate grading of tumor is vital in determination of the appropriate treatment. This is a retrospective study on 12 T2-weighted (T2W) images (n=12) of patients with pathologically diagnosed glioma of different grades retrieved at the Radiology Department, Hospital Universiti Sains Malaysia from Picture Archiving and Communications System (PACS). A single slice of T2W image is chosen for analysis and all of the images were analysed for its image quality. Then, a workflow and protocol for image processing were explored using ImageJ. The lesion and normal appearing white matter (NAWM) region of interest (ROI) were selected for histogram analysis, along with lesion to normal tissue ratio (LNR) calculations and standard deviation of lesion (SDL) analysis for the assessment of tumor heterogeneity and intensity. Additionally, color thresholding, lookup table (LUT) images with 3D plot surface images, and midline shift angle measurements were used to assess tumor characteristics such as the margin, edema, and mass effect. The result of this study shows no specific trend for the tumor intensity based on LNR findings; however the SDL of the tumor shows an increasing trend across the glioma grades, which proves the increasing heterogeneity as the glioma grades increase. The margin of the tumor can be depicted by entropy thresholding, the edema is depicted by LUT images. Finally, the mass effect is depicted by the measurement of midline shift angle where higher-grade tumor depicts further deviation from 180.0° and in 3D surface plot images. In conclusion, ImageJ-based radiomic analysis provides an accessible and simple method for classifying glioma brain tumors. This approach may potentially facilitate tumor grading without the use of GBCAs.
Description
Keywords
-