QP-Prostate CAD
Radiological CADe/CADx software that automatically detects and identifies suspected lesions for clinically significant prostate cancer (csPCa; Gleason ≥7) on bi-parametric prostate MRI (T2w + DWI). Outputs include 3D prostate gland segmentation and DICOM overlay markings of suspected lesions in two suspicion levels (moderate/high). For physicians qualified to read and interpret prostate MRI per ACR PI-RADS recommendations. Patients above 40 years.
Regulatory
| Regulatory status | 510(k) cleared |
| FDA submission pathway | 510(k) |
| FDA decision date | 2025-03-18 |
| FDA 510(k) number | K242683 |
| CE marked | Yes |
Validation summary
| Study design | Retrospective |
| Cohort size | 228 patients; 247 samples — Pivotal dataset: 228 cases collected retrospectively from multiple US centers, completely independent from training data. Standalone lesion-level analysis included N=247 lesion observations (positives + negatives from targeted biopsies plus negative cases without biopsy). MRMC: same 228 cases × 9 readers, fully crossed factorial design. |
| Number of validation sites | — |
| Site geography | Multi-center (US) |
| Comparator | Predicate device |
| Primary endpoint | AUC |
| Primary result | MRMC primary endpoint: AUCunaided 0.849 (95% CI 0.814, 0.884); AUCaided 0.868 (95% CI 0.834, 0.902); ΔAUC +0.019 (95% CI 0.001, 0.038), p=0.039 — statistically significant. Standalone (lesion level): AUC-ROC 0.732 (95% CI 0.668, 0.791); sensitivity at high-suspicion marker 0.677 (95% CI 0.593, 0.761) with FPR/case 0.417; sensitivity at high+moderate markers 0.795 (95% CI 0.722, 0.861) with FPR/case 0.855. |
| External validation | Yes — Pivotal dataset (n=228) acquired in US at multiple centers, independent from training institutions. Demographics: age range 43–81, mean 64.7; 85.1% White, 8.3% African American, 1.3% Asian, 1.8% Other, 3.5% declined/unavailable; 3.1% Hispanic; 86.0% on 3T scanners (14.0% 1.5T); Siemens 62.3%, Philips 20.2%, GE 17.5%. · Subgroup analyses showed consistent performance across age (≤65, >65), BMI (<25, ≥25), race/ethnicity, and scanner vendor; minor performance deviation on 1.5T attributed to inherently lower image quality vs 3T (3T preferable when available). |
| Peer-reviewed | No |
| FDA decision summary | View on accessdata.fda.gov |
Limitations noted
[Source-detail preserved: Study design: Multi-center; Primary endpoint (original): AUC_aided_vs_unaided] (i) Pivotal dataset enriched 50:50 positive:negative — not population prevalence; (ii) standalone AUC 0.732 is modest; high-+moderate-marker operating point has high FPR/case 0.855; (iii) 1.5T subgroup underperformed vs 3T; (iv) no peer-reviewed pivotal publication identified — only FDA Decision Summary and Quibim press releases.
Deployment
| Available | Yes |
Sources
- https://www.accessdata.fda.gov/cdrh_docs/pdf24/K242683.pdf · fda_db · 2026-05-12
- https://quibim.com/qp-prostate/ · company · 2026-05-12
Elbakri A. OncologyAI Registry: QP-Prostate CAD. Available at: https://oncologyairegistry.org/tools/quibim-qp-prostate.html. Accessed [date].