Resumen:
La inteligencia artificial (IA) ha sido propuesta como una herramienta para evaluar la densidad mamográfica del tejido mamario (MBD). Este estudio tuvo como objetivo evaluar el acuerdo en la clasificación de MBD entre cuatro radiólogos (lectores humanos [HRs]) con diferentes años de experiencia en imágenes mamarias y la IA Lunit INSIGHT MMG. Este estudio transversal se llevó a cabo con una muestra de conveniencia de radiólogos capacitados en imágenes mamarias que evaluaron mamografías de detección de MBD de mujeres asintomáticas de 35 años o más utilizando descriptores BI-RADS. El kappa de Cohen determinó el acuerdo entre los HRs y la IA. Se incluyeron un total de 192 mujeres con una edad media de 55.4 ± 31.8 años (rango de 37 a 82 años). El acuerdo interobservador entre los HRs y la IA varió en la Categoría a, pero fue sustancial en la Categoría b (HR1 k = 0.729, HR2 k = 0.718, HR3 k = 0.768 y HR4 k = 0.672) y en la Categoría c, HR1, HR2 y HR3 tuvieron un acuerdo sustancial (k = 0.728, k = 0.697 y k = 0.738, respectivamente) y HR4 tuvo un acuerdo moderado (k = 0.578), mientras que en la Categoría d, fue mayormente moderado. Los acuerdos entre los HRs y la IA variaron de justo a sustancial. Los HRs con más años de experiencia en la interpretación de imágenes mamarias tuvieron un menor acuerdo con la IA para la clasificación de MBD que los HRs con menos tiempo de experiencia.

Abstract:
Artificial intelligence (AI) has been proposed as a tool for assessing mammographic breast density (MBD). This study aimed to evaluate the agreement of MBD classification between four radiologists (human readers [HRs]) with different years of experience in breast imaging and the AI Lunit INSIGHT MMG. This cross-sectional study was conducted with a convenience sample of radiologists trained in breast imaging who assessed MBD screening mammograms of asymptomatic women 35 years or older using BI-RADS descriptors. Cohen’s kappa determined the agreement between the HRs and AI. A total of 192 women with a mean age of 55.4 ± 31.8 years (range 37-82 years) were included. Interobserver agreement between HRs and AI varied in Category a but was substantial in Category b (HR1 k = 0.729, HR2 k = 0.718, HR3 k = 0.768, and HR4 k = 0.672) and in Category c, HR1, HR2, and HR3 had substantial agreement (k = 0.728, k = 0.697, and k = 0.738, respectively) and HR4 had moderate agreement (k = 0.578), while in Category d, it was mostly moderate. HRs and AI agreements varied from fair to substantial. HRs with more years of experience in breast image interpretation had a lower agreement with AI for MBD classification than HRs with less time.

Abstract:
Artificial intelligence (AI) has been proposed as a tool for assessing mammographic breast density (MBD). This study aimed to evaluate the agreement of MBD classification between four radiologists (human readers [HRs]) with different years of experience in breast imaging and the AI Lunit INSIGHT MMG. This cross-sectional study was conducted with a convenience sample of radiologists trained in breast imaging who assessed MBD screening mammograms of asymptomatic women 35 years or older using BI-RADS descriptors. Cohen’s kappa determined the agreement between the HRs and AI. A total of 192 women with a mean age of 55.4 ± 31.8 years (range 37-82 years) were included. Interobserver agreement between HRs and AI varied in Category a but was substantial in Category b (HR1 k = 0.729, HR2 k = 0.718, HR3 k = 0.768, and HR4 k = 0.672) and in Category c, HR1, HR2, and HR3 had substantial agreement (k = 0.728, k = 0.697, and k = 0.738, respectively) and HR4 had moderate agreement (k = 0.578), while in Category d, it was mostly moderate. HRs and AI agreements varied from fair to substantial. HRs with more years of experience in breast image interpretation had a lower agreement with AI for MBD classification than HRs with less time.