Premium
SU‐G‐JeP2‐07: Fusion Optimization of Multi‐Contrast MRI Scans for MR‐Based Treatment Planning
Author(s) -
Zhang L,
Yin F,
Liang X,
Cai J
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4957027
Subject(s) - voxel , nuclear medicine , contrast (vision) , contrast to noise ratio , image registration , image fusion , weighting , artificial intelligence , radiation treatment planning , medicine , computer science , medical imaging , mathematics , radiology , image (mathematics) , radiation therapy , image quality
Purpose: To develop an image fusion method using multiple contrast MRI scans for MR‐based treatment planning. Methods: T1 weighted (T1‐w), T2 weighted (T2‐w) and diffusion weighted images (DWI) were acquired from liver cancer patient with breath‐holding. Image fade correction and deformable image registration were performed using VelocityAI (Varian Medical Systems, CA). Registered images were normalized to mean voxel intensity for each image dataset. Contrast to noise ratio (CNR) between tumor and liver was quantified. Tumor area was defined as the GTV contoured by physicians. Normal liver area with equivalent dimension was used as background. Noise was defined by the standard deviation of voxel intensities in the same liver area. Linear weightings were applied to T1‐w, T2‐w and DWI images to generate composite image and CNR was calculated for each composite image. Optimization process were performed to achieve different clinical goals. Results: With a goal of maximizing tumor contrast, the composite image achieved a 7–12 fold increase in tumor CNR (142.8 vs. −2.3, 11.4 and 20.6 for T1‐w, T2‐w and DWI only, respectively), while anatomical details were largely invisible. With a weighting combination of 100%, −10% and −10%, respectively, tumor contrast was enhanced from −2.3 to −5.4, while the anatomical details were clear. With a weighting combination of 25%, 20% and 55%, balanced tumor contrast and anatomy was achieved. Conclusion: We have investigated the feasibility of performing image fusion optimization on multiple contrast MRI images. This mechanism could help utilize multiple contrast MRI scans to potentially facilitate future MR‐based treatment planning.