Search Results

Genius AI Detection is a computer-aided detection and diagnosis (CADe/CADx) software device intended to be used with compatible digital breast tomosynthesis (DBT) systems to identify and mark regions of interest including soft tissue densities (masses, architectural distortions and asymmetries) and calcifications in DBT exams from compatible DBT systems and provide confidence scores that offer assessment for Certainty of Findings and a Case Score. The device intends to aid in the interpretation of digital breast tomosynthesis exams in a concurrent fashion, where the interpreting physician confirms or dismisses the findings during the reading of the exam.

Genius Al Detection is a software device intended to identify potential abnormalities in breast tomosynthesis images. Genius Al Detection analyzes each standard mammographic view in a digital breast tomosynthesis examination using deep learning networks. For each detected lesion, Genius Al Detection produces CAD results that include the location of the lesion, an outline of the lesion and a confidence score for that lesion. Genius Al Detection also produces a case score for the entire tomosynthesis exam.

Genius Al Detection packages all CAD findings derived from the corresponding analysis of a tomosynthesis exam into a DICOM Mammography CAD SR object and distributes it for display on DICOM compliant review workstations. The interpreting physician will have access to the CAD findings concurrently to the reading of the tomosynthesis exam. In addition, a combination of peripheral information such as number of marks and case scores may be used on the review workstation to enhance the interpreting physician's workflow by offering a better organization of the patient worklist.

The acceptance criteria for the Genius AI Detection device and the study that proves it meets these criteria can be summarized as follows:

1. Table of Acceptance Criteria and Reported Device Performance:

2. Sample Size for Test Set and Data Provenance:

• Sample Size: 764 cases for standalone testing, including 106 cancers and 658 non-cancer cases. For the MRMC study, 390 cases were used (106 cancers and 284 negative cases).
• Data Provenance: Not explicitly stated in the provided text, but it's implied that the data is from Digital Breast Tomosynthesis (DBT) exams. No country of origin or whether it was retrospective or prospective is mentioned.

3. Number of Experts and Qualifications for Ground Truth:

• Number of Experts: 17 readers participated in the MRMC study.
• Qualifications of Experts: Described as "MQSA-Qualified Interpreting Physicians and Radiologists." No specific years of experience or subspecialty focus are detailed.

4. Adjudication Method for Test Set:

• The document implies that the ground truth for cases with cancer was established prior to the reader study (e.g., "106 cancers"). However, the specific method of adjudication (e.g., 2+1, 3+1 expert consensus, pathology, follow-up) for establishing the definitive ground truth for the test set (both cancer and non-cancer cases) is not explicitly described in the provided text. It only states that the MRMC study utilized "106 cancers, and 284 negative cases."

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study:

• Yes, an MRMC comparative effectiveness study was conducted.
• Effect Size of Human Readers with AI vs. without AI Assistance:
  • AUC Improvement: The average observed AUC increased by +0.031 (95% Cl: 0.012, 0.051) with CAD.
  • Sensitivity Improvement: The average observed reader sensitivity for cancer cases increased by +9.0% (99% Cl: 6.0%, 12.1%) with CAD.
  • Recall Rate Change: The average observed recall rate for non-cancer cases increased by +2.4% (99% Cl: 0.7%, 4.2%) with CAD.
  • Read-time Change: The average observed case read-time increased by 5.7s (95% Cl: 4.9s to 6.4s) with CAD.

6. Standalone (Algorithm Only) Performance Study:

• Yes, a standalone study was conducted.
• Purpose: To establish equivalence of Genius AI Detection performance on Hologic's standard resolution tomosynthesis images (~100um) compared to its performance on high-resolution tomosynthesis images (70μm).
• Methodology: The standalone study was conducted on paired high-resolution and standard-resolution 3D data sets, both acquired from a single exposure under the same compression.
• Results: The study confirmed "comparable detection performance" as observed by fROC analysis, with "no significant differences... in either acquisition mode," including when stratified by lesion type and breast density.

7. Type of Ground Truth Used:

• The text frequently refers to "cancers" and "non-cancer cases." This implies the ground truth for cancer diagnoses was established through pathology (biopsy), which is the definitive method for cancer. For "non-cancer cases" it would likely be a combination of imaging findings confirmed by follow-up or expert consensus, though this is not explicitly detailed.

8. Sample Size for Training Set:

• The sample size for the training set is not provided in the given document.

9. How Ground Truth for Training Set was Established:

• How the ground truth for the training set was established is not provided in the given document.

Ask a specific question about this device