Evaluating primary tumor position variation for rectal cancer patients treated with long course radiotherapy

TL;DR

This study compares two methods to track tumor movement in rectal cancer patients during radiotherapy, finding that surface-based methods detect more variation than center-of-mass approaches.

Contribution

The study introduces a surface-based metric to quantify tumor positional variations during radiotherapy, revealing limitations of the COM method.

Findings

Systematic and random errors increased with surface-based metrics compared to COM measurements.

Higher-located tumors showed larger left-right and AP surface variations.

Surface variations suggest COM-based analysis may underestimate deformations.

Abstract

Objective. To quantify interfraction shape and positional variations of primary tumor volumes for rectal cancer patients receiving long course radiotherapy by comparing two quantification strategies: a center-of-mass (COM) method and a surface-based metric that captures local deformations. Approach. This study utilized repeat MRI scans before and during radiotherapy (RT) for rectal cancer to investigate the positional variation of the primary gross tumor volume (GTVp). Sixteen patients underwent six MRI exams, with the initial three before the RT course and the subsequent three at one, two, and four weeks into the RT course. GTVp’s were delineated on 3D T2-weighted MRIs, and positional variation analyzed using both COM and point-based surface displacements against the initial scan. Surface displacements were quantified using a bidirectional local distance measure, analyzing 3D displacement vectors. Additionally, the study examined local right–left (RL) and anterior–posterior (AP) surface variations relative to tumor height in the rectum by mapping baseline GTVp volumes onto a reference rectum structure. Main results. Systematic error for COM measurements were 1.7, 1.3 and 2.0 mm for AP, RL, and cranial–caudal (CC) direction, respectively. Random errors were 2.1, 1.2 and 2.2 mm, while the GM errors were −0.3, 0.5 and −0.3 mm for AP, RL, and CC directions, respectively. An increase in systematic and random errors were observed when comparing 95th percentile surface displacements to the COM measurements, indicating local displacements which the COM did not detect. Additionally, a general tendency for higher-located tumors to experience larger left–right and AP surface variations were seen when evaluating the 95th percentile. Significance. COM-based analysis might underestimate local deformations. Consequently, surface-based methods might provide more robust estimations of systematic, random and group mean errors for planning target volume-margin calculation. The surface variations tend to increase for tumors located in the upper part of the rectum.

Figures