Análisis de métricas de calidad de imagen en equipos de radiología digital dentro de un programa de control de calidad remoto y automatizada
Fecha
2024
Tipo
tesis de maestría
Autores
Soto Chinchilla, Yerry Alberto
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
Se realizó un estudio cuyo propósito fue investigar la correlación entre métricas de calidad de imagen para equipos digitales de mamografía y radiología, para ello se creó una página web que almacenara las métricas y mostrara los cambios en el tiempo mediante gráficos de control, asimismo, se utilizaron rangos intercuartílicos para determinar si hubo valores que salieron de una banda de dos desviaciones estándar alrededor de la media en una distribución normal. Las métricas de calidad de imagen se extrajeron de un total de 163 imágenes para radiología convencional desde el 6 de junio de 2017 hasta el 25 de marzo de 2021 abarcando un antes y después de cambiar el tubo de rayos X y 144 de mamografía desde el 5 de junio de 2017 hasta el 26 de junio de 2020 abarcando un cambio de detector en el mamógrafo. También se obtuvo de dichas imágenes sus mapas de varianza. Los resultados muestran que las métricas que arrojan diferencias significativas para el caso de radiología convencional fueron la Relación Señal Ruido (SNR) e índice de sensibilidad (d') y para mamografía se evidencian dichas diferencias en las métricas Relación Diferencia de Señal Ruido (SDNR), Relación Señal Ruido e Índice de Sensibilidad (d'), además en virtud del análisis temporal de los mapas de varianza se comprobó que en mamografía existía fatiga del equipo que pueden provocar la existencia de artefactos que se manifiesten en el futuro que podrían llegar a ser significativos en imágenes médicas cuando se requiera la búsqueda de pequeños objetos.
The study was aimed to investigate the correlation between image quality metrics for digital mammography and radiology equipment, for this, a web page was created that would store the metrics and show the changes over the time using control charts.In addition, interquartile ranges were used to determine if there were values that fell outside a band of two standard deviations around the mean in a normal distribution. The image quality metrics were extracted from a total of 163 conventional radiology images from June 6th, 2017 to March 25th, 2021, covering a period before and after changing the X-ray tube, and 144 mammography images from June 5th, 2017 to June 26th, 2020, covering a detector change in the mammography machine. Their variance maps were also obtained from these images. The results indicate that there are metrics that show significant differences for the case of conventional radiology were Signal to Noise Ratio (SNR) and sensitivity index (d'), for mammography these differences are evidenced in the metrics Signal Difference to Noise Ratio (SDNR), Signal to Noise Ratio and sensitivity index (d'), furthermore, based on the temporal analysis of variance maps, it was verified that in mammography, there was equipment fatigue that could lead to the presence of artifacts in the future. These artifacts might become significant in medical images when the search for small objects is necessary.
The study was aimed to investigate the correlation between image quality metrics for digital mammography and radiology equipment, for this, a web page was created that would store the metrics and show the changes over the time using control charts.In addition, interquartile ranges were used to determine if there were values that fell outside a band of two standard deviations around the mean in a normal distribution. The image quality metrics were extracted from a total of 163 conventional radiology images from June 6th, 2017 to March 25th, 2021, covering a period before and after changing the X-ray tube, and 144 mammography images from June 5th, 2017 to June 26th, 2020, covering a detector change in the mammography machine. Their variance maps were also obtained from these images. The results indicate that there are metrics that show significant differences for the case of conventional radiology were Signal to Noise Ratio (SNR) and sensitivity index (d'), for mammography these differences are evidenced in the metrics Signal Difference to Noise Ratio (SDNR), Signal to Noise Ratio and sensitivity index (d'), furthermore, based on the temporal analysis of variance maps, it was verified that in mammography, there was equipment fatigue that could lead to the presence of artifacts in the future. These artifacts might become significant in medical images when the search for small objects is necessary.
Descripción
Palabras clave
RADIOLOGÍA, CONTROL DE CALIDAD, MAMOGRAFÍA, CALIDAD DE IMAGEN, MÉTRICAS