Diagnostic performance of S-detect artificial intelligence compared with conventional breast ultrasound for lesion characterization: a prospective lesion-based study

Abstract Purpose The aim of this prospective study was to compare the diagnostic performance of conventional breast ultrasound (US) assessment with S-Detect artificial intelligence (AI) for differentiating benign from malignant breast lesions. The study was not designed to assess reduction in biopsy rates. Materials and Methods A total of 130 women with 130 target breast lesions were prospectively evaluated from January 2024 to December 2025. One target lesion was analyzed per patient. Each lesion underwent conventional B-mode US with final BI-RADS assessment and S-Detect analysis. Final histopathology was used as the reference standard. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated with 95% confidence intervals (CIs), and paired classifications were compared using McNemar’s test. Results Histopathology showed 109 benign lesions (83.9%) and 21 malignant lesions (16.1%). Conventional US showed 90.5% sensitivity, 87.2% specificity, 57.6% PPV, 97.9% NPV, and 87.7% accuracy. S-Detect showed the same sensitivity (90.5%) but lower specificity (63.3%), PPV (32.2%), and accuracy (67.7%), with an NPV of 97.2%. S-Detect produced 40 false-positive results compared with 14 for conventional US. On paired comparison, S-Detect generated significantly more positive classifications (McNemar P < 0.001). Conclusion S-Detect showed sensitivity comparable to conventional US but lower specificity, PPV, and overall accuracy, with more false-positive results. Therefore, the current findings do not demonstrate a reduction in unnecessary biopsy. S-Detect may be used as an adjunct to conventional US assessment, but its clinical role requires further evaluation in larger prospective reader studies using predefined management rules.

Authors

Institutions

Publication Details

Journal
The Egyptian Journal of Radiology and Nuclear Medicine
Published
2026-11-14
DOI
https://doi.org/10.1186/s43055-026-01910-0
Primary Topic
AI in cancer detection
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Diagnostic performance of S-detect artificial intelligence compared with conventional breast ultrasound for lesion characterization: a prospective lesion-based study

Manal Mohamed Refaat Beshir, Ghada Mohamed, Amira Hamed Radwan, Fifi Samy Mahmoud et al.
The Egyptian Journal of Radiology and Nuclear Medicine
AI in cancer detection
article

Diagnostic performance of S-detect artificial intelligence compared with conventional breast ultrasound for lesion characterization: a prospective lesion-based study

Manal Mohamed Refaat Beshir, Ghada Mohamed, Amira Hamed Radwan, Fifi Samy Mahmoud, Amr Farouk Moustafa
article en

Abstract

Abstract Purpose The aim of this prospective study was to compare the diagnostic performance of conventional breast ultrasound (US) assessment with S-Detect artificial intelligence (AI) for differentiating benign from malignant breast lesions. The study was not designed to assess reduction in biopsy rates. Materials and Methods A total of 130 women with 130 target breast lesions were prospectively evaluated from January 2024 to December 2025. One target lesion was analyzed per patient. Each lesion underwent conventional B-mode US with final BI-RADS assessment and S-Detect analysis. Final histopathology was used as the reference standard. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated with 95% confidence intervals (CIs), and paired classifications were compared using McNemar’s test. Results Histopathology showed 109 benign lesions (83.9%) and 21 malignant lesions (16.1%). Conventional US showed 90.5% sensitivity, 87.2% specificity, 57.6% PPV, 97.9% NPV, and 87.7% accuracy. S-Detect showed the same sensitivity (90.5%) but lower specificity (63.3%), PPV (32.2%), and accuracy (67.7%), with an NPV of 97.2%. S-Detect produced 40 false-positive results compared with 14 for conventional US. On paired comparison, S-Detect generated significantly more positive classifications (McNemar P < 0.001). Conclusion S-Detect showed sensitivity comparable to conventional US but lower specificity, PPV, and overall accuracy, with more false-positive results. Therefore, the current findings do not demonstrate a reduction in unnecessary biopsy. S-Detect may be used as an adjunct to conventional US assessment, but its clinical role requires further evaluation in larger prospective reader studies using predefined management rules.

The Egyptian Journal of Radiology and Nuclear MedicineVol. 57(1)
Cairo University (EG)
AI in cancer detection
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon