Does Correction factor based T10 atlas help to achieve quick and improved Ktrans computation? a feasibility study

Abstract

Introduction

The ultrafast dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) used for Pharmacokinetic (PK) analysis is an effective method to predict benign and malignant tumor. Attempts have been made to achieve an accurate K^trans by correcting T₁₀, a major factor that influence K^trans computation.

Purpose

The purpose of this study to develop a correction factor atlas database at every spatial-location (x,y coordinate) in both side of the mMR breast coil cuff space to improve T₁₀ homogeneity in the reference method using multiple tube phantom by SW based automation.

Method

Both multiple tube phantoms (61 tubes in each phantom) and patient's studies were acquired on simultaneous PET/MRI Biograph mMR system (Siemens, Erlangen, Germany) by using 4 channel mMR breast coil. Each spatial location correction factors (2916 each side, total 4392 correction factors) derived from multiple tube phantom were assign to an inhouse developed software (Radvista, version 3.2.95) that compute K^trans in few seconds. A retrospective analysis of DCE-MRI data acquired for 60 seconds of 146 patients with mean age of 50 years (24-80 years) having 178 enhancing histologically proved breast lesions forms the material of the study.

Results

Corrected and non-corrected ROC curve analysis revealed a mean K^trans, v_e value of 1.52 min^-1, 0.44 & 1.16 min^-1, 0.34 respectively. The sensitivity, specificity and overall accuracy of K^trans for non-corrected data were 89.06%, 74.00% and 84.83% respectively and for corrected data it was 89.84%, 86.00%, and 88.76% respectively. The AUC of corrected data was improved to 0.927 (95% CI 0.878 to 0.960) from 0.893 (95% CI 0.839 to 0.935) for non-corrected data.

Conclusion

Spatially normalized T₁₀ values derived from correction factor based T₁₀ atlas in a reference method can provide a quick estimation of K^trans with improved diagnostic accuracy.