Quantifying Systematic Bias In Caliper-Based Tumor Volume Measurements Using Micro-CT
Abstract
Purpose
Murine models of head and neck squamous cell carcinoma (HNSCC) are routinely monitored using external caliper measurements to estimate tumor volume, most often relying on simplified geometric assumptions such as cylindrical approximations. While convenient, these approximations do not reflect the heterogenous morphology and shape of tumors. Therefore, in this study, we directly compared caliper-derived tumor volumes with day-matched volumetric measurements obtained from micro–computed tomography (μCT).
Methods
Orthotopic tumors were induced with injection of MOC1 (n=42), MLM1(n=39) or MOC2 (n=36) tumor cells into the right buccal mucosa of C57BL/6 mice. When tumors reached ~50mm3, they were treated with 2 fractions of 8 Gy radiation with 5mg/kg cisplatin (days 0 and 7). At day 14, mice underwent μPET/CT imaging to assess tumor volume, and caliper measurements were performed. Tumor volumes from caliper measurements were calculated using a cylindrical approximation. Tumor volumes from μPET/CT were calculated by contouring the volume in three dimensions via Hounsfield thresholding (values 75-300 HU) in the region of buccal asymmetry. Volumes derived from CT were considered ground truth, as they encompassed the irregular tumor shape.
Results
Linear regression analysis demonstrated a strong but non-unity relationship between the two measurement approaches. The fitted slope was 0.72, indicating that caliper-based methods systematically underestimate tumor volume relative to CT-derived measurements. This consistent bias enables applying a systematic correction factor to improve the accuracy of caliper-derived tumor volume estimates.
Conclusion
This study highlights limitations in simple geometric approximations of tumor volume and emphasizes the importance of incorporating image-based volumetric analysis when precise tumor burden quantification is required, such as studies evaluating treatment response.