Background: Nowadays, mammography and DCE-MRI are the gold standard for breast cancer screening. Scientific experience demonstrate an increasing application of Echo elastosonography in breast cancer diagnosis, for this reason, within the Synergy-Net project the objective is to create a system based on machine learning algorithms, a CAD developed with CNNs, capable of representing a decision support in the analysis of echo-elastic sonographic images. Results: 315 female patients at the “Vanvitelli” Breast Unit were subjected to digital 3D mammography tomosynthesis and advanced ultrasound (quantitative Echo elastosonography) and comparative analysis of the two methods. We collected 70 breast cancer, 200 benign pathologies (dropout of 11.1%) and we compared two different homogeneous by composition subgroups (35 with cancer and 100 with benign pathology 20.5%), relating to the diagnostic predictive capabilities of 3D digital mammography and quantitative elastosonography: Group A age 40–50 years, Group B age 50–60 years. Results demonstrate fair predictive performances of echoelastosonography versus traditional mammography, especially in the first group (40–50 years), p < 0.05 in favor of elastosonography regarding sensitivity and accuracy; performances that appear substantially comparable in the second group (50–60 years). We also tested the diagnostic predictive capabilities of the CAD Synergy-Breast-Net achieving encouraging results: Sensitivity 80%, Specificity 72%, Accuracy 74%, Negative Predictive Value 84.7% and Positive Predictive Value 83.3%. Conclusion: From this perspective, the screening of patients aged < 40 years and integrate the main screening activities in patients aged between 40 and 50 years integrated with increasingly high-performance machine learning systems, could represent a valid alternative in the future.
Artificial Intelligence in Breast Cancer Diagnosis: “Synergy-Net” in Campania FESR-POR (European Fund of Regional Development—Regional Operative Program) Research Project
Parmeggiani D.;Luongo P.;Allaria A.;Ruggiero R.;Sciarra A.;Parisi S.;Fisone F.;Donnarumma M. C.;Della Monica P.;Di Domenico M.;Docimo L.;Agresti M.
2024
Abstract
Background: Nowadays, mammography and DCE-MRI are the gold standard for breast cancer screening. Scientific experience demonstrate an increasing application of Echo elastosonography in breast cancer diagnosis, for this reason, within the Synergy-Net project the objective is to create a system based on machine learning algorithms, a CAD developed with CNNs, capable of representing a decision support in the analysis of echo-elastic sonographic images. Results: 315 female patients at the “Vanvitelli” Breast Unit were subjected to digital 3D mammography tomosynthesis and advanced ultrasound (quantitative Echo elastosonography) and comparative analysis of the two methods. We collected 70 breast cancer, 200 benign pathologies (dropout of 11.1%) and we compared two different homogeneous by composition subgroups (35 with cancer and 100 with benign pathology 20.5%), relating to the diagnostic predictive capabilities of 3D digital mammography and quantitative elastosonography: Group A age 40–50 years, Group B age 50–60 years. Results demonstrate fair predictive performances of echoelastosonography versus traditional mammography, especially in the first group (40–50 years), p < 0.05 in favor of elastosonography regarding sensitivity and accuracy; performances that appear substantially comparable in the second group (50–60 years). We also tested the diagnostic predictive capabilities of the CAD Synergy-Breast-Net achieving encouraging results: Sensitivity 80%, Specificity 72%, Accuracy 74%, Negative Predictive Value 84.7% and Positive Predictive Value 83.3%. Conclusion: From this perspective, the screening of patients aged < 40 years and integrate the main screening activities in patients aged between 40 and 50 years integrated with increasingly high-performance machine learning systems, could represent a valid alternative in the future.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
