RadioLens v1.0

RadioLens v1.0 is a medical image management and processing system featuring an optional AI-driven module, SpindleX, for the quantitative analysis of spinal stress X-rays in adults to detect vertebral displacements and assess ligamentous injuries. The system's methodology relies on semi-automated anatomical identification and calculation of specific spinal parameters, with performance validation demonstrating high sensitivity for image characterization and strong agreement with radiologists for critical measurements.

RadioLens v1.0 is a medical diagnostic application designed for displaying, processing, storing, and transferring DICOM data, with the exception of mammography. It provides capabilities for image filtering, digital manipulation, and quantitative measurements, operating on standard personal and business computers. A significant component is the optional SpindleX module, an artificial intelligence (AI) tool that analyzes cervical and lumbar digital stress X-rays. SpindleX assists qualified medical practitioners by semi-automatically identifying relevant spine anatomy and calculating vertebral displacement measurements. These measurements are then used to determine the severity and location of potential spinal ligament injuries leading to subluxation, utilizing criteria published in the AMA Guides to The Evaluation of Permanent Impairment, specifically for adult patients (>18 years) who have not undergone prior surgery or experienced other traumatic injuries causing spinal anatomy distortion. The SpindleX AI module was trained using a substantial dataset of 13,452 cervical and 6,492 lumbar de-identified retrospective studies from over 200 U.S. institutions. Its performance was rigorously evaluated in a multi-reader multi-center (MRMC) retrospective study with 600 de-identified stress X-rays (300 cervical, 300 lumbar), independent of the training data, involving three expert radiologists. The SpindleX module demonstrated 100% sensitivity for body part identification (cervical vs. lumbar) and high sensitivities for patient position identification (flexion: 96.8%, extension: 92.0%, neutral: 90.5%), meeting predefined targets. For quantitative measurements such as stress lines, canal diameter, vertebral offset, and motion segment integrity, Intraclass Correlation Coefficient (ICC) scores indicated that the agreement between the radiologists and the model was largely comparable to the agreement among the radiologists themselves, with most cervical and lumbar measurements achieving "Good" to "Excellent" levels of agreement.

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 the SpindleX module compare when analyzing X-rays from different equipment manufacturers or with varying image acquisition protocols?; What is the processing time for the SpindleX module per case, and how does it integrate into existing radiological workflows?; Are there plans to expand the SpindleX module's capabilities to analyze other spinal regions or different types of spinal pathologies beyond ligamentous injury and subluxation?; How does the SpindleX algorithm account for anatomical variations or mild degenerative changes that might mimic or confound signs of acute ligamentous injury in the adult population?; What were the specific inter-rater reliability scores (ICC) among the three expert radiologists alone for each measured parameter in the validation study, to provide a baseline for the 'Rads & Model ICC' scores?