RESUMEN:
Objetivo: El objetivo de este trabajo fue evaluar los resultados de un software basado en un algoritmo de inteligencia artificial para predicción del riesgo de malignidad en nódulos mamarios. Materiales y Métodos: Estudio retrospectivo y unicéntrico que evaluó 555 nódulos mamarios sometidos a biopsia percutánea en un centro de referencia oncológica. Los hallazgos ultrasonográficos fueron clasificados de acuerdo con el léxico del BI-RADS. Las imágenes fueron analizadas por el software Koios DS Breast y divididas en benignas o probablemente benignas, sospecha baja o intermedia, sospecha alta o probablemente malignas. El resultado histopatológico fue considerado como patrón oro. Resultados: La media de edad de las pacientes fue de 51 años y el tamaño medio de los nódulos fue de 16 mm. La sensibilidad y la especificidad fueron del 99,1% y 34,0% para el radiólogo y del 98,2% y 39,0% para el software, respectivamente. El valor predictivo positivo para malignidad para las categorías BIRADS fue similar para el radiólogo y para el software. Se identificaron dos resultados falso-negativos en la evaluación por el radiólogo que fueron clasificados como sospechosos por el software, y cuatro resultados falso-negativos en la evaluación por el software que fueron clasificados como sospechosos por el radiólogo. Conclusión: En nuestra muestra, el software de inteligencia artificial demostró resultados comparables a la evaluación por el radiólogo.

ABSTRACT:
Objective: To evaluate the results obtained with an artificial intelligence-based software for predicting the risk of malignancy in breast masses from ultrasound images. Materials and Methods: This was a retrospective, single-center study evaluating 555 breast masses submitted to percutaneous biopsy at a cancer referral center. Ultrasonographic findings were classified in accordance with the BI-RADS lexicon. The images were analyzed by using Koios DS Breast software and classified as benign, probably benign, low to intermediate suspicion, high suspicion, or probably malignant. The histological classification was considered the reference standard. Results: The mean age of the patients was 51 years, and the mean mass size was 16 mm. The radiologist evaluation had a sensitivity and specificity of 99.1% and 34.0%, respectively, compared with 98.2% and 39.0%, respectively, for the software evaluation. The positive predictive value for malignancy for the BI-RADS categories was similar between the radiologist and software evaluations. Two false-negative results were identified in the radiologist evaluation, the masses in question being classified as suspicious by the software, whereas four false-negative results were identified in the software evaluation, the masses in question being classified as suspicious by the radiologist. Conclusion: In our sample, the performance of artificial intelligence-based software was comparable to that of a radiologist.

ABSTRACT:
Objective: To evaluate the results obtained with an artificial intelligence-based software for predicting the risk of malignancy in breast masses from ultrasound images. Materials and Methods: This was a retrospective, single-center study evaluating 555 breast masses submitted to percutaneous biopsy at a cancer referral center. Ultrasonographic findings were classified in accordance with the BI-RADS lexicon. The images were analyzed by using Koios DS Breast software and classified as benign, probably benign, low to intermediate suspicion, high suspicion, or probably malignant. The histological classification was considered the reference standard. Results: The mean age of the patients was 51 years, and the mean mass size was 16 mm. The radiologist evaluation had a sensitivity and specificity of 99.1% and 34.0%, respectively, compared with 98.2% and 39.0%, respectively, for the software evaluation. The positive predictive value for malignancy for the BI-RADS categories was similar between the radiologist and software evaluations. Two false-negative results were identified in the radiologist evaluation, the masses in question being classified as suspicious by the software, whereas four false-negative results were identified in the software evaluation, the masses in question being classified as suspicious by the radiologist. Conclusion: In our sample, the performance of artificial intelligence-based software was comparable to that of a radiologist.