TRAQinform IQ

TRAQinform IQ is a software device that provides adjunctive quantitative analysis of lesions from PET/CT scans to aid clinicians in assessing changes over time. It utilizes machine learning for segmentation and automated algorithms to track Regions of Interest, quantifying changes in volume, density, and tracer uptake to generate a detailed report for review by medical professionals.

TRAQinform IQ is a software-only device designed to provide quantitative analysis of lesions, identified as Regions of Interest (ROI), from previously acquired PET/CT DICOM-compliant imaging data. It serves as an adjunct to local image interpretation, with reports generated by the device manufacturer and signed by a U.S. board-certified physician. The system employs automated processes including machine learning for skeletal and anatomical segmentation, threshold-based ROI identification, and automated matching of ROIs between multiple CT and PET/CT images. It performs quantitative analysis to assess changes in ROI volume, density (for CT), and tracer uptake (specifically Fludeoxyglucose F18 for multi-timepoint analysis, but also PSMA and SSTR-targeted drugs for single-timepoint), calculating various metrics like SUV values, Hounsfield Units, and tracking complex ROI dynamics. Performance evaluation of TRAQinform IQ included component testing, a test-retest study establishing limits of repeatability for imaging features, and a pivotal reader study. In the reader study with 103 patients, oncologists used the TRAQinform Report adjunctively, leading to reevaluation of progression status in 32 patients. When these cases were assessed by an expert panel, patient-level overall percent agreement for progression status was 41%-76% (95% CI), with a positive percent agreement of 78% and a negative percent agreement of 36%. At the ROI level, for ROIs highlighted due to the TRAQinform Report, the panel's agreement with the device's classification of ROI changes (new, increasing, unchanged, decreasing, disappearing) ranged from 63% to 90%.

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. Given the observed discrepancies between TRAQinform IQ and local PET/CT reporting, particularly the 36% Negative Percent Agreement in the pivotal study, what specific image characteristics or patient conditions contribute most to these differences and how might the algorithm be refined to improve negative predictive value?; How does the machine learning model for skeletal and anatomic structure segmentation handle anatomical variations or post-surgical changes, and what is its performance robustness across different scanner types and imaging protocols not extensively covered in the provided data?; The pivotal study mentions reevaluation prompted by the TRAQinform Report. What were the primary quantitative metrics or ROI change classifications from the report that most frequently led oncologists to alter their initial progression assessment?; For multi-timepoint analysis beyond FDG, such as with PSMA or SSTR-targeted PET drugs (currently indicated for single-timepoint), what modifications or additional validation would be required for TRAQinform IQ to reliably quantify longitudinal changes?; Considering the TRAQinform Report is generated by the device manufacturer, what is the typical turnaround time for report generation, and how does this workflow integrate into time-sensitive clinical decision-making processes, especially when discrepancies necessitate multidisciplinary review?