Inteligencia artificial para la detección temprana del cáncer de mama por mamografía
Fecha
2023
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad de Guayaquil. Facultad de Ciencias Médicas. Escuela de Graduados
Resumen
Antecedentes: La inteligencia artificial (IA) es una rama de la informática dedicada al desarrollo
de algoritmos informáticos que emulan la inteligencia humana. Los avances tecnológicos en IA
han dado lugar a técnicas que podrían aumentar la detección del cáncer de mama, así como
mejorar la eficiencia clínica en las prácticas de diagnóstico por imágenes y la toma de decisiones
con respecto a estrategias de detección y prevención.
Objetivo: Establecer la eficacia de la inteligencia artificial para la detección temprana del cáncer
de mama por mamografía.
Metodología: Estudio correlacional, no experimental, transversal, retrospectivo. Se incluyeron
169 pacientes con mamografías realizadas en el Hospital Guasmo Sur de enero a diciembre del
2021. Se usó el software de IA (Lunit Insight MMG) y reportes médicos. El análisis estadístico
se realizó programa SPSS versión 26.
Resultados: La composición de mama de predominio fue fibroglandular dispersa, y se obtuvo
una correlación entre IA y el radiólogo (rho Sperman=0,86; p < 0.05). Entre Score de anormalidad
y Bi-rads agrupado existe una asociación moderada (V de Cramer = 0,548; p < 0.05). Curvas ROC
para la IA fue AUC 0,808 (p < 0.05, IC: 0,64 a 0,97; S:66.7%; E:95%) y para el Bi-rads AUC
0,737 (p < 0.05, IC: 0,64 a 0,97; S:50%; E: 92%) interpretando que la IA es un buen test y el
reporte del radiólogo es un test regular; así la IA para la detección de cáncer de mama fue
(S:62%;E:98%).
Conclusión: La inteligencia artificial demostró su eficacia para diagnóstico de cáncer de mama
Background. Artificial intelligence (AI) is a branch of computer science dedicated to the development of computer algorithms that emulate human intelligence. Technological advances in AI have given rise to techniques that could increase breast cancer detection, as well as improve clinical efficiency in diagnostic imaging practices and decision-making regarding detection and prevention strategies. Objective: To establish the efficacy of artificial intelligence for the early detection of breast cancer by mammography. Methodology: Correlational, non-experimental, cross-sectional, retrospective study. 169 patients with mammograms performed at Guasmo Sur Hospital from January to December 2021 were included. AI software (Lunit Insight MMG) and medical reports were used. Statistical analysis was performed with SPSS version 26 program. Results: The predominant breast composition was dispersed fibroglandular, and a correlation was obtained between IA and the radiologist (Sperman's rho=0.86; p < 0.05). There is a moderate association between abnormality score and grouped Bi-rads (Cramer's V = 0.548; p < 0.05). ROC curves for IA was AUC 0.808 (p < 0.05, CI: 0.64 to 0.97; S:66.7%; E:95%) and for Bi-rads AUC 0.737 (p < 0.05, CI: 0, 64 to 0.97; S: 50%; E: 92%) interpreting that the AI is a good test and the radiologist's report is a regular test; thus, the AI for the detection of breast cancer was (S:62%;E:98%). Conclusion: Artificial intelligence demonstrated its effectiveness for the diagnosis of breast cancer
Background. Artificial intelligence (AI) is a branch of computer science dedicated to the development of computer algorithms that emulate human intelligence. Technological advances in AI have given rise to techniques that could increase breast cancer detection, as well as improve clinical efficiency in diagnostic imaging practices and decision-making regarding detection and prevention strategies. Objective: To establish the efficacy of artificial intelligence for the early detection of breast cancer by mammography. Methodology: Correlational, non-experimental, cross-sectional, retrospective study. 169 patients with mammograms performed at Guasmo Sur Hospital from January to December 2021 were included. AI software (Lunit Insight MMG) and medical reports were used. Statistical analysis was performed with SPSS version 26 program. Results: The predominant breast composition was dispersed fibroglandular, and a correlation was obtained between IA and the radiologist (Sperman's rho=0.86; p < 0.05). There is a moderate association between abnormality score and grouped Bi-rads (Cramer's V = 0.548; p < 0.05). ROC curves for IA was AUC 0.808 (p < 0.05, CI: 0.64 to 0.97; S:66.7%; E:95%) and for Bi-rads AUC 0.737 (p < 0.05, CI: 0, 64 to 0.97; S: 50%; E: 92%) interpreting that the AI is a good test and the radiologist's report is a regular test; thus, the AI for the detection of breast cancer was (S:62%;E:98%). Conclusion: Artificial intelligence demonstrated its effectiveness for the diagnosis of breast cancer
Descripción
PDF
Palabras clave
NEOPLASMAS DE LA MAMA, RAYOS X, MAMOGRAFIA, HOSPITAL GUASMO SUR, CANTON GUAYAQUIL, ECUADOR