The PI-QUAL scoring sheet (Fig. 1) was created to standardise the assessment of prostate mpMRI quality and streamline the collection of data both for clinical and research purposes.
All patients included in this report gave written informed consent to have their images used for research and teaching purposes.
PI-QUAL workflow
After a general window in which the radiologist is asked to enter their name and the name of the scan site (if the program is used for research purposes), the PI-QUAL software automatically extrapolates the technical parameters of T2-weighted imaging, diffusion-weighted imaging and dynamic contrast-enhanced sequences from the raw Digital Imaging and Communications in Medicine (DICOM) images as outlined in the PI-QUAL scoring sheet (Fig. 1) and checks their compliance against the PI-RADS technical requirements. It should be noted that the PI-QUAL software has been built following PI-RADS v. 2.1 guidelines [1].
The radiologist then manually evaluates the scans for the presence (or absence) of the items listed in the ‘visual assessment’ box of the scoring sheet. Finally, the radiologist uses both results to state whether the images for each sequence are of diagnostic quality.
In the final step, the operator inserts the PI-QUAL score and has also the possibility to include additional comments and relevant snapshots from the different sequences (Figs. 2, 3 and 4).
Technical aspects
When the DICOM data is imported, the software parses the header and stores the elements of this information in its metadata repository for the user’s reference when browsing the patient list. The header contents of DICOM datasets are also stored in the memory for retrieval in other software functions. The PI-QUAL workflow is configured to directly read the values of specific tags embedded in the DICOM data from memory storage, then the workflow returns the values of these tags as variables for display to the user or for further manipulation in the workflow.
For example, the slice thickness of the axial T2-weighted image is pulled from the DICOM tag, then the workflow compares this value to the PI-QUAL minimum standard value (as per PI-RADS v. 2.1 guidelines) for slice thickness. The workflow displays a table to the user that contains the image-specific value, the value of the PI-QUAL minimum standard and a boolean pass (yes)/fail (no) result when comparing the image-specific value against the minimum standard value. The workflow gives the user the ability to adjust the pass (yes)/fail (no) results for any DICOM-based technical parameters and perform other visual-based assessments for the image that cannot be directly derived from DICOM information (e.g., clear delineation of certain anatomical structures).
Therefore, the overall quality scores are based upon the radiologists’ interpretations of both the automatic technical parameter results and the visual assessment according to the PI-RADS v. 2.1 guidelines.
Image analysis
In this study, two specialist consultant radiologists (C.A. and F.G. reporting more than 3,000 and 2,000 prostate MRI scans per year, respectively) analysed in consensus the image quality of 10 multiparametric scans (all without endorectal coil) from different MRI systems and vendors. First, they filled in the PI-QUAL scoring sheet manually and then, after an interval of 8 weeks between the two readings to avoid any recall bias, they re-assessed the image quality using a dedicated PI-QUAL software program (MIM® Symphony Dx v. 7.1.2 - Cleveland, OH, USA).
The time needed to assess the image quality for each scan was recorded for both methods.
Statistical analysis
Data are presented as medians and interquartile ranges (IQR) and were compared using a two-tailed Wilcoxon test. All statistical analyses were performed by using SPSS (version 27.0; SPSS, Chicago, IL, USA); p values were considered to indicate a significant difference when < 0.05.