Precision Diffusion-Weighted Imaging of Head and Neck Squamous Cell Carcinoma.

Abstract

Background and purpose: Head and neck squamous cell carcinoma (HNSCC) represents a significant global health burden, with over 890,000 new cases annually and approximately 450,000 related deaths worldwide. Our study aims to optimize diffusion-weighted imaging (DWI) sequences for the head and neck and determine reproducibility metrics values for ADC, allowing clinicians to distinguish between genuine physiological changes and measurement errors.

Materials and methods: We optimized the head and neck DWI protocol using QIBA-recommended parameters, high 2000 b-value, Zoom-IT technology, and a Polyvinylpyrrolidone (PVP) phantom calibration. Twenty-six biopsy-proven HNSCC patients underwent pre-treatment "coffee-break" test-retest DWI scans on a Siemens 3T Vida MRI scanner. ADC values were extracted from 46/51 segmented tumor regions (either tumor mass or metastatic lymphadenopathy), and reproducibility measures were assessed using the within-subject coefficient of variation (wCV), repeatability coefficient (RC), and intraclass correlation coefficient (ICC). Bland-Altman plots and correlation analyses were used to evaluate measurement precision. Both manual segmentation by two expert neuroradiologists and semiautomated methods of segmentation were utilized.

Results: Out of 51 identified tumor masses, 46 were included in the final analysis. The Bland-Altman plot revealed minimal bias in ADC values between test-retest scans. The calculated 5.2-5.4% wCV resulted in a repeatability coefficient of 14.4%. The ICC between test and retest measurements was 0.8-0.83, indicating good test-retest reliability. The DICE score between the two expert readers for all parameters indicated excellent agreement between segmentations performed by different readers, with values of 0.991 for the test measurement and 0.976 for the retest measurement.

Conclusions: Our findings support the use of an optimized DWI protocol for accurate and reproducible ADC quantification in HNSCC, which is comparable to or better than the percent repeatability coefficient derived from QIBA Profiles for breast, liver, and prostate, although it is inferior to those for the brain. Further QIB Stage 2-type Profile research is necessary to refine imaging parameters, improve reproducibility markers, and establish DWI as a clinical standard for evaluating HNSCC tumor recurrence.

Abbreviations: BOT=base of tongue. HNC=head and neck cancer. HNSCC= head and neck squamous cell carcinoma. ICC=Intraclass correlation coefficient. QIBA=Qualitative Initiative Biomarker Alliance. QuiC=Quantitative Imaging Committee. LoA= Limits of agreement. RC=repeatability coefficient. SCC=Squamous cell carcinoma. QIB=quantitative imaging biomarkers. wCV=within-lesion coefficient of variation.