Implications of the partial volume effect correction on the spatial quantification of hypoxia based on [18F]FMISO PET/CT data

Abstract

Purpose

This study evaluates the impact of partial volume effect (PVE) correction on [¹⁸F]fluoromisonidazole (FMISO) PET images, focusing on the conversion of standardized uptake values (SUV) to partial oxygen pressure (pO₂) and the subsequent determination of hypoxic tumor volume (HTV).

Methods

FMISO PET images from 49 head and neck squamous cell carcinoma cases were retrospectively corrected for PVE and converted to pO₂. A pO₂ threshold of 10 mmHg was used to delineate the HTV (HTV_pO2). Comparisons of pO₂ distribution and HTV_pO2 between corrected and uncorrected images were made, with pO₂ distributions evaluated against published polarographic data. HTV_pO2 was compared to HTV defined by the conventional tumor-to-muscle ratio (TMR) method (HTV_TMR) in terms of volume and topography (DICE coefficient, Hausdorff distance, and center-of-gravity distance) across different TMR cutoff levels. The cutoff level where the segmentation results from both methods were most similar was identified (TMR_best).

Results

The PVE correction led to decreased minimum pO₂, increased HTV_pO2 and the identification of more hypoxic cases (HTV > 0). The pO₂ distribution demonstrated improved alignment with published polarographic data. At TMR_best 1.6, the center-of-gravity distance between HTV_TMR and HTV_pO2 demonstrated a low median at 1.5 mm, while the wide range (0.0 to 9.6 mm) indicated high interpatient variability. The shape of HTV exhibited considerable variation with DICE 0.74 (0.03 to 1.00) and Hausdorff distance 8.5 mm (2.0 to 42.8 mm).

Conclusions

PVE correction is recommended before converting SUV to pO₂ for the spatially resolved quantification of hypoxia.