QP-Brain®

QP-Brain® is a software application that utilizes automated image processing techniques, including artificial intelligence, to analyze brain MRI scans for quantitative volumetric measurements of brain structures and white matter hyperintensities. The system aims to provide clinicians with objective data from 3D T1-weighted and T2 FLAIR MR images to support the assessment of structural brain changes in adult and adolescent patients, with outputs comparable to normative data and suitable for longitudinal tracking.

QP-Brain® is a medical image processing software designed to assist clinicians in assessing structural MRIs for adults and adolescents aged 18 and older. It automates the labeling and volumetric quantification of segmentable brain structures using 3D T1-weighted MRI scans, and identifies white matter hyperintensities (WMH) from T2 FLAIR images, employing artificial intelligence for WMH detection. The system's workflow involves checking MR image compatibility, then an automated pipeline performs artifact correction, segmentation, atlas-based parcellation, volume calculation, and generates reports. The outputs, including segmented images and quantitative data, are compatible with DICOM workstations and PACS and can be compared against normative data if patient age and gender are provided, and it allows for longitudinal information reporting. The performance of QP-Brain® was assessed by comparing its automated segmentations to manual expert segmentations as a reference standard. For volumetric analysis of brain structures such as Gray Matter, White Matter, Cerebrospinal Fluid, and Intracranial Volume from T1 MRIs, the software showed high agreement, with DICE scores generally above 0.95 (e.g., GM 0.983, WM 0.990) and relatively small volume differences. In the quantification of White Matter Hyperintensities from T2 FLAIR images, performance, measured by DICE scores and absolute volume error, varied depending on the WMH load, with DICE scores ranging from approximately 0.4-0.5 for low WMH to over 0.8 for very high WMH. The software also demonstrated a Pearson’s correlation coefficient of 0.88 for WMH count.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the performance of QP-Brain® in quantifying white matter hyperintensities (WMH) vary across different MRI scanner manufacturers, field strengths, and acquisition protocols?; What specific artificial intelligence algorithms are employed for WMH segmentation, how were they trained and validated, and what is their robustness to atypical WMH presentations or image artifacts?; Could you elaborate on the demographic characteristics, size, and source of the normative database used for percentile comparisons, and how it accounts for variables like ethnicity or education level?; For longitudinal analysis, what are the specific image registration and processing techniques used to ensure accurate and reliable quantification of changes over time, especially with varying scan intervals or slight protocol differences?; Beyond volumetric data, does QP-Brain® offer any information on the spatial distribution or morphological characteristics of segmented brain structures or WMH, which could be clinically relevant?