MRI information of the prostate is increasingly used for improving the diagnostic yield of prostate biopsies . However, increasing complexity of a procedure makes it prone to errors at multiple technical and human levels. Incorporating MRI information and ultrasonography (US) images for MRI-fusion biopsies is a technically challenging task. It involves various steps such as the acquisition and fusion of MRI and US images, the needle guidance during biopsy, and the diligence of the pathological evaluation of biopsy specimens. These different steps and interfaces between different medical professions influence the diagnostic performance of MRI-fusion biopsies.
For example, in daily clinical practice, MRIs from different institutions still harbour a great variance of sequences and reporting, despite the European Society of Urogenital Urology (ESUR) recently introducing acquisition and imaging protocols and a new and advanced version of the Prostate Imaging Reporting and Data System (PIRADS) version 2.0 . The usefulness of such reporting schemes is evidenced by a moderate-to-good interobserver agreement between uro-radiologists for tumour lesion interpretation and corresponding κ values ranging from 0.55 to 0.80 . Important pitfalls of image interpretation are benign lesions such as prostatitis, BPH and fibrosis, which might score similarly to prostate cancer lesions. This problem is further aggravated by a high proportion of patients that receive their first multiparametric MRI (mpMRI) of the prostate in the repeat-biopsy setting with a high burden of post-biopsy artefacts (haemorrhage, capsular irregularity) and lower overall cancer detection rate. Also, during MRI-fusion biopsy patient movement, prostate deformation by the US probe, and mismatch of image planes can lead to a biopsy error exceeding 4 mm. Moreover, targeting error might be aggravated by MRI underestimation of the tumour volume compared with final pathology . After various authors reported the advantages and accuracy of MRI/US-fusion biopsy approaches, Cash et al.  address potential reasons for targeted biopsy failure to detect prostate cancer compared with random biopsy. Within their analyses the authors address potential limitations and technical considerations. Based on different technical biopsy strategies (with the patient placed within the MRI scanner (‘in-bore’) vs outside) and different technical approaches, these considerations are very important.
In contrast to cognitive fusion, most MRI/US platforms allow needle tracking by archiving the needle orientation, either by an electromagnetic, image-based or stepper-based mechanism . However, lesion targeting by needle guidance is highly dependent on the dimensions of the primary lesion, numbers of relevant lesions, localisation, and overall prostate volume, making MRI-US fusion and cognitive fusion more error prone (i.e. aiming off the mark with the needle) than in-bore biopsies. Moreover, different technical fusion approaches provide different degrees of manual/automated adjustment tools, with for example either rigid or elastic image transformation to facilitate MRI/US image alignment.
In their analyses, Cash et al.  found that 34% of negative targeted biopsies could be explained by initially too high estimated PIRADS scores that were downgraded at re-reading. Interestingly, the remaining lesions were without an mpMRI correlate but within this group 92.9% showed a primary Gleason 3 pattern in biopsy pathology, suggesting a high degree of invisibility on mpMRI. Subanalyses did not show an association of targeted biopsy failures in the ventral location. Therefore, the study by Cash et al.  is an important precursor for further analyses to address other underlying reasons for targeted biopsy failure. Moreover, it reveals the need for a tight collaboration of radiologists, urologists, and pathologists as interdisciplinary partners involved in MRI-fusion biopsy. Consequently, the optimal diagnostic performance of MRI-fusion biopsies can only be achieved through standardised MRI performance, reading and reporting of MRI findings, as well as final correlation of MRI findings with histopathological work up.