Transpara

Transpara™ is a radiological computer-assisted detection and diagnosis software designed to aid physicians in interpreting screening mammograms by identifying suspicious breast lesions and assessing their likelihood of malignancy. Utilizing machine learning algorithms trained on extensive databases and DICOM integration, the software analyzes mammograms to provide CAD marks, region-specific scores, and an overall exam score, with version 1.5.0 demonstrating improved detection performance and expanded vendor compatibility.

Transpara™ is a software-only medical device designed to assist physicians in the interpretation of screening mammograms for breast cancer detection and diagnosis. Its primary purpose is to serve as a concurrent reading aid, identifying regions suspicious for breast cancer, such as soft tissue lesions and calcifications, and assessing their likelihood of malignancy. The system provides output in several forms: computer-aided detection (CAD) marks on suspicious areas, region-based scores displayed upon user query, and an overall exam score (Transpara™ Score from 1-10) indicating the exam's likelihood of containing cancer. The intended users are qualified physicians reading screening mammograms, and the patient population is women undergoing such screenings; however, patient management decisions should not be based solely on Transpara™'s analysis. The Transpara™ system operates as a DICOM node, receiving images from mammography devices or PACS archives, and employs a processing server and an optional viewer. Its core technology involves machine learning algorithms trained on large, biopsy-proven datasets of cancerous, benign, and normal breast tissue to recognize suspicious findings. The software pre-processes images to ensure compatibility across different mammography vendor equipment, allowing consistent analysis. Key user interface functions include activating CAD marks (with a focus on low false positives for soft tissue lesions), querying specific regions for a suspiciousness level, and viewing the exam-based Transpara™ Score. Non-clinical performance testing of Transpara™ version 1.5.0 demonstrated improved detection performance compared to its predicate (version 1.3.0) for multiple mammography manufacturers and confirmed similar performance for newly supported Fujifilm modalities, indicating its effectiveness in detecting lesions.

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 'exam based Transpara™ Score' (1-10) quantitatively correlate with established risk stratification models or BI-RADS categories, and what was the specific methodology used for validating this scoring system's clinical utility?; What were the precise sensitivity, specificity, and area under the ROC curve (AUC) metrics from the validation testing that demonstrate Transpara™ 1.5.0's 'better detection performance' compared to version 1.3.0, and what were the absolute and relative improvements observed for different lesion types (calcifications vs. soft tissue masses)?; Can you provide more technical details on the image pre-processing techniques that enable Transpara™ to analyze images from different vendors using the same core algorithms, and how is the robustness of these techniques validated against variations in image acquisition protocols?; The system aims for a 'very low false positive rate' for soft tissue lesion marks. What is the quantified false positive rate per image or per exam for these marks, how does it compare to the false positive rate for calcification marks, and what is the trade-off with sensitivity for these soft tissue lesions?; Given that the pivotal reader study was conducted with the predicate device (Transpara 1.3.0), what specific bridging data or analysis was provided to extrapolate the safety and effectiveness findings to Transpara™ 1.5.0, particularly concerning the algorithmic improvements and expanded modality support?