The impact of the co-registration technique and analysis methodology in comparison studies between advanced imaging modalities and whole-mount-histology reference in primary prostate cancer

• Main Authors: Constantinos Zamboglou, Maria Kramer, Selina Kiefer, Peter Bronsert, Lara Ceci, August Sigle, Wolfgang Schultze-Seemann, Cordula A. Jilg, Tanja Sprave, Thomas F. Fassbender, Nils H. Nicolay, Juri Ruf, Matthias Benndorf, Anca L. Grosu, Simon K. B. Spohn
• Format: Article
• Language: English
• Published: Nature Publishing Group 2021-03-01
• Series: Scientific Reports
• Online Access: https://doi.org/10.1038/s41598-021-85028-5

Abstract Comparison studies using histopathology as standard of reference enable a validation of the diagnostic performance of imaging methods. This study analysed (1) the impact of different image-histopathology co-registration pathways, (2) the impact of the applied data analysis method and (3) intraindividually compared multiparametric magnet resonance tomography (mpMRI) and prostate specific membrane antigen positron emission tomography (PSMA-PET) by using the different approaches. Ten patients with primary PCa who underwent mpMRI and [18F]PSMA-1007 PET/CT followed by prostatectomy were prospectively enrolled. We demonstrate that the choice of the intermediate registration step [(1) via ex-vivo CT or (2) mpMRI] does not significantly affect the performance of the registration framework. Comparison of analysis methods revealed that methods using high spatial resolutions e.g. quadrant-based slice-by-slice analysis are beneficial for a differentiated analysis of performance, compared to methods with a lower resolution (segment-based analysis with 6 or 18 segments and lesions-based analysis). Furthermore, PSMA-PET outperformed mpMRI for intraprostatic PCa detection in terms of sensitivity (median %: 83–85 vs. 60–69, p < 0.04) with similar specificity (median %: 74–93.8 vs. 100) using both registration pathways. To conclude, the choice of an intermediate registration pathway does not significantly affect registration performance, analysis methods with high spatial resolution are preferable and PSMA-PET outperformed mpMRI in terms of sensitivity in our cohort.