BriefCase-Quantification

BriefCase-Quantification is an AI-powered software designed to automate the measurement of brain midline shift from non-contrast head CT images, aiding medical specialists by providing quantitative data and longitudinal tracking. Its methodology relies on a deep learning algorithm to process DICOM images and generate reports, with clinical studies demonstrating high accuracy (0.94 mm mean absolute error) and minimal bias compared to manual neuroradiologist assessments.

BriefCase-Quantification is an AI-driven radiological image management and processing software designed for the automatic measurement of brain midline shift (MLS) in non-contrast head CT (NCCT) images. It is intended for use in adults and transitional adolescents (18 years and older) to assist trained medical specialists by automating the traditionally manual MLS measurement process. The system operates on a cloud-based Linux server, processing filtered DICOM images using a deep learning algorithm. It can analyze images from a single time point or perform comparative analysis of images from the same individual acquired at multiple time points, without altering the original medical data. The output, delivered to PACS or a desktop application, includes a summary report with a preview image of the slice showing the largest MLS, annotations of the midline and septum pellucidum distance, and a table or graph for longitudinal MLS tracking. The software's performance was validated in a retrospective, multicenter, blinded study involving 284 NCCT cases from 228 unique patients, compared against a ground truth established by three neuroradiologists. The primary endpoint, the mean absolute error between the algorithm's measurement and the ground truth, was 0.94 mm (95% CI: 0.74 mm, 1.14 mm), meeting prespecified goals. A secondary endpoint, evaluated using a Bland-Altman plot, showed a mean difference of -0.15 mm, indicating minimal bias. For analyses of multiple time points for the same patient, the mean absolute error was 1.16 mm for initial scans and 1.28 mm for follow-up scans, also meeting performance goals. The device is intended as an assistive tool and not for stand-alone diagnostic decision-making; clinicians maintain responsibility for comprehensive image review.

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 algorithm's performance vary across different CT scanner manufacturers and imaging protocols beyond those tested in the pivotal study?; What is the typical processing time for a single NCCT series, and how does this integrate into high-volume clinical workflows?; Are there specific anatomical variants or pathological conditions (e.g., severe skull fractures, post-craniotomy changes) that might affect the accuracy of the midline shift measurement?; How does the software handle cases with suboptimal image quality or significant motion artifacts?; What were the specific prespecified performance goals for the primary and secondary endpoints in the pivotal study, and how do they compare to clinically acceptable margins of error for MLS measurement?