WE-G-17A-02: MRI-Based Lung Tumor Tracking with Navigator Echo Pulses

Authors


Abstract

Purpose:

To evaluate the feasibility of directly tracking the displacement of lung tumors in MRI using navigator echo pulses, and to compare the navigator-derived tumor positions with manually-identified tumor positions.

Methods:

Twelve patients with lung tumors underwent free-breathing sagittal MR scans under an IRB-approved protocol for 8–12 min. We used a multi-slice 2D TruFISP pulse sequence on a 1.5T Siemens Avanto scanner. Scan parameters were as follows: TE=1.29ms, TR=2.57ms, flip angle=60°, pixel matrix=176×256, resolution=1.95mm × 1.95 mm × (9–16) mm, acquisition time=2.5s. One-dimensional profiles of the images in the SI and AP directions that passed through the center of the tumor were selected for analysis as simulated navigator echo profiles. The tumor position was identified through edge detection and tracked throughout the temporally varying images in each direction. The navigator-tracked direct tumor positions were compared with tumor centroid positions manually identified by an expert physician on the full images (200 acquisitions per scan). The correlation coefficient and pixel distance between the two tumor positions were calculated. In addition, respiratory surrogate navigator positions of the diaphragm and chest wall were also evaluated for correlation with the manually-identified tumor positions.

Results:

Direct tracking of the lung tumor position with navigator echoes was more highly correlated to the manual tumor positions than surrogate tracking in 72% of cases with significant correlations (p<0.005). Only four patients exhibited strong linear correlations (R2>0.6). The average distance between the navigator tumor positions and the manually-identified tumor positions was 0.99±0.74 pixels (1.93±1.44 mm).

Conclusions:

Lung tumor displacement can be tracked using navigator echoes to within 2 mm (1 pixel) of manually identified tumor positions.

This work was supported in part by Grant No. CA124766 from the NIH/NCI, and Varian Medical Systems (Palo Alto, CA).

Ancillary