# Multi-institution longitudinal apparent diffusion coefficient measurements in a diffusion weighted imaging phantom at room temperature

## TL;DR

This study shows that measuring ADC values in cancer imaging is reliable across multiple hospitals and scanners without complex setup.

## Contribution

Demonstrated reliable ADC measurement repeatability and reproducibility across multiple institutions using a room-temperature phantom.

## Key findings

- Mean ADC bias across all scanners and sessions was less than 0.01 × 10−3 mm² s−1 (0.81%).
- Reproducibility was 0.07 × 10−3 mm² s−1 (9%), meeting QIBA requirements.
- Results support the feasibility of multi-institution longitudinal ADC studies with minimal harmonization.

## Abstract

•ADC was measured on six MR scanners at four institutions over 18 months.•A room temperature DWI phantom was used.•95 % limits of agreement for ADC bias for all scans was 0.07 × 10−3 mm2 s−1 (9 %).•We found good ADC accuracy, short and long-term repeatability and reproducibility.•All scanners met the QIBA requirements for ADC bias, error and repeatability.

ADC was measured on six MR scanners at four institutions over 18 months.

A room temperature DWI phantom was used.

95 % limits of agreement for ADC bias for all scans was 0.07 × 10−3 mm2 s−1 (9 %).

We found good ADC accuracy, short and long-term repeatability and reproducibility.

All scanners met the QIBA requirements for ADC bias, error and repeatability.

This work contributes to technical validation of apparent diffusion coefficient (ADC) as a biomarker of cancer. The aim was to evaluate ADC accuracy, random error, short-term and long-term repeatability and reproducibility, across multiple institutions using a room temperature phantom.

ADC measurements were made in a travelling room temperature diffusion weighted imaging (DWI) phantom on six scanners at four UK institutions over 18 months at six-month intervals. ADC bias measurements were calculated as the difference between measured and temperature corrected ground-truth ADC values and used to calculate mean ADC bias, isocentre ADC error estimate, short- and long-term intra-scanner repeatability as per the Quantitative Imaging Biomarkers Alliance (QIBA) DWI profiles, and inter-scanner reproducibility by calculating the 95 % limits of agreement for all ADC bias measurements.

The use of a room-temperature phantom with a magnetic resonance (MR) readable thermometer enabled ADC measurements without ice-water setup, considerably simplifying logistics with respect to multi-institution ADC quality assurance. Mean ADC bias across all scanners and sessions was <0.01 × 10−3 mm2 s−1 (0.81 %); mean isocentre ADC error estimate was 1.43 %; average scanner short-term repeatability was <0.01 × 10−3 mm2 s−1 (1 %). Reproducibility was 0.07 × 10−3 mm2 s−1 (9 %).

Results indicated good ADC accuracy, repeatability and reproducibility; demonstrating the feasibility of transferring diagnostic DWI sequences between scanners from the same manufacturer, for use in multi-institution longitudinal studies, and assessing ADC with minimal quality control and harmonisation steps required.