Cover Image

Aplicação do Sonobreast e comparação de Acurácia e concordância com o sistema BI-RADS em nódulos mamários sólidos ao ultrassonografia

Saved in:
Bibliographic Details
Main Author: Oliveira, Luís Fernando Pádua
Publication Date: 2014
Format: Doctoral thesis
Language: por
Source: Repositório Institucional da UFG
Download full: http://repositorio.bc.ufg.br/tede/handle/tede/7765
Summary: Objectives: To compare the accuracy and concordance of SONOBREAST with the BI-RADS system for predicting malignancy in solid breast masses observed on ultrasound; and application the SONOBREAST in women with breast lumps. Methods: 274 women were included prospectively, with the assessment of 500 breast masses observed on ultrasound and that presented any surgical indication. Sonographic features were assessed, the patients' age and family history of first degree. The likelihood of malignancy according to SONOBREAST model was calculated through the webpage <www.sonobreast.com.br>, and the BI-RADS system. The result was compared to anatomopathological examination. Results: The average age of patients was 41.92 (± 14.40) years. There were 87 (17.40%) malignant tumors; and 382 (76.40%) nonpalpable lesions. For the cutoff of 2%, the lesions were regarded as suspicious in 171 cases (34.20%) by both SONOBREAST and BI-RADS. The Kappa coefficient of concordance between the methods was 1 (p <0.001). SONOBREAST was identical to the BI-RADS system for sensitivity (95.40)%, specificity (78.69%), positive predictive value (48.54%), negative predictive value (98.78%) and accuracy (81.60 %). Compared to categorical variables (3, 4 and 5) classified by BIRADS, the area under the ROC curve by SONOBREAST was of 94.41 (92.20 to 96.62) and by BI-RADS of 89.99 (86.60 to 93.37). Conclusion: The SONOBREAST model presents identical accuracy compared to the BIRADS system, when using the same parameters for the cutoff for malignancy suspicion. Considering the increasing likelihood of malignancy for categories 3, 4 and 5 of the BI-RADS system, SONOBREAST model allowed a more precise and individualized assessment.
Summary:Objectives: To compare the accuracy and concordance of SONOBREAST with the BI-RADS system for predicting malignancy in solid breast masses observed on ultrasound; and application the SONOBREAST in women with breast lumps. Methods: 274 women were included prospectively, with the assessment of 500 breast masses observed on ultrasound and that presented any surgical indication. Sonographic features were assessed, the patients' age and family history of first degree. The likelihood of malignancy according to SONOBREAST model was calculated through the webpage <www.sonobreast.com.br>, and the BI-RADS system. The result was compared to anatomopathological examination. Results: The average age of patients was 41.92 (± 14.40) years. There were 87 (17.40%) malignant tumors; and 382 (76.40%) nonpalpable lesions. For the cutoff of 2%, the lesions were regarded as suspicious in 171 cases (34.20%) by both SONOBREAST and BI-RADS. The Kappa coefficient of concordance between the methods was 1 (p <0.001). SONOBREAST was identical to the BI-RADS system for sensitivity (95.40)%, specificity (78.69%), positive predictive value (48.54%), negative predictive value (98.78%) and accuracy (81.60 %). Compared to categorical variables (3, 4 and 5) classified by BIRADS, the area under the ROC curve by SONOBREAST was of 94.41 (92.20 to 96.62) and by BI-RADS of 89.99 (86.60 to 93.37). Conclusion: The SONOBREAST model presents identical accuracy compared to the BIRADS system, when using the same parameters for the cutoff for malignancy suspicion. Considering the increasing likelihood of malignancy for categories 3, 4 and 5 of the BI-RADS system, SONOBREAST model allowed a more precise and individualized assessment.

Similar Items