Evaluating a radiotherapy deep learning synthetic CT algorithm for PET-MR attenuation correction in the pelvis

期刊论文
出版模式：

开放获取
发表日期：

2024年1月29日

查看全文

Jonathan J. Wyatt,
Sandeep Kaushik,
Cristina Cozzini,
Rachel A. Pearson,
George Petrides,
Florian Wiesinger,
Hazel M. McCallum,
Ross J. Maxwell

摘要

Background

Positron emission tomography–magnetic resonance (PET-MR) attenuation correction is challenging because the MR signal does not represent tissue density and conventional MR sequences cannot image bone. A novel zero echo time (ZTE) MR sequence has been previously developed which generates signal from cortical bone with images acquired in 65 s. This has been combined with a deep learning model to generate a synthetic computed tomography (sCT) for MR-only radiotherapy. This study aimed to evaluate this algorithm for PET-MR attenuation correction in the pelvis.

Methods

Ten patients being treated with ano-rectal radiotherapy received a $^{18}$ $^{18}$ F-FDG-PET-MR in the radiotherapy position. Attenuation maps were generated from ZTE-based sCT (sCTAC) and the standard vendor-supplied MRAC. The radiotherapy planning CT scan was rigidly registered and cropped to generate a gold standard attenuation map (CTAC). PET images were reconstructed using each attenuation map and compared for standard uptake value (SUV) measurement, automatic thresholded gross tumour volume (GTV) delineation and GTV metabolic parameter measurement. The last was assessed for clinical equivalence to CTAC using two one-sided paired t tests with a significance level corrected for multiple testing of $p \le 0.05/7 = 0.007$ $p \leq 0.05 / 7 = 0.007$ . Equivalence margins of $\pm 3.5\%$ $\pm 3.5 %$ were used.

Results

Mean whole-image SUV differences were −0.02% (sCTAC) compared to −3.0% (MRAC), with larger differences in the bone regions (−0.5% to −16.3%). There was no difference in thresholded GTVs, with Dice similarity coefficients $\ge 0.987$ $\geq 0.987$ . However, there were larger differences in GTV metabolic parameters. Mean differences to CTAC in ${\mathrm {SUV}}_{\max}$ ${SUV}_{max}$ were $1.0 \pm 0.8\%$ $1.0 \pm 0.8 %$ (± standard error, sCTAC) and $-4.6 \pm 0.9\%$ $- 4.6 \pm 0.9 %$ (MRAC), and $1.0 \pm 0.7\%$ $1.0 \pm 0.7 %$ (sCTAC) and $-4.3 \pm 0.8\%$ $- 4.3 \pm 0.8 %$ (MRAC) in ${\mathrm {SUV}}_{\rm mean}$ ${SUV}_{mean}$ . The sCTAC was statistically equivalent to CTAC within a $\pm 3.5\%$ $\pm 3.5 %$ equivalence margin for ${\mathrm {SUV}}_{\max}$ ${SUV}_{max}$ and ${\mathrm {SUV}}_{\rm mean}$ ${SUV}_{mean}$ ( $p = 0.007$ $p = 0.007$ and $p = 0.002$ $p = 0.002$ ), whereas the MRAC was not ( $p = 0.88$ $p = 0.88$ and $p = 0.83$ $p = 0.83$ ).

Conclusion

Attenuation correction using this radiotherapy ZTE-based sCT algorithm was substantially more accurate than current MRAC methods with only a 40 s increase in MR acquisition time. This did not impact tumour delineation but did significantly improve the accuracy of whole-image and tumour SUV measurements, which were clinically equivalent to CTAC. This suggests PET images reconstructed with sCTAC would enable accurate quantitative PET images to be acquired on a PET-MR scanner.