Isabela developed the research in partnership with the Laboratory for Translational Data Science team

Dr. ISABELA PANZERI CARLOTTI BUZATTO presented at the Brazilian Breast Cancer Symposium 2023 the results of her study conducted with the partnership Laboratory for Translational Data Science entitled Axillary ultrasound and fine-needle aspiration cytology to predict clinically relevant nodal burden in breast cancer patients where it was observed that Machine learning can reliably predict the malignancy of BI-RADS; 4a and 4b breast lesions based on clinical and ultrasonographic features. The work was awarded with the José Aristódemo Pinotti Award.

MACHINE LEARNING CAN RELIABLY PREDICT MALIGNANCY OF BI-RADS 4A AND 4B BREAST LESIONS BASED ON CLINICAL AND ULTRASONOGRAPHIC FEATURES

Isabela Panzeri Carlotti Buzatto¹, Daniel Guimarães Tiezzi¹, Sarah Abud Recife², Ruth Morais Bonini³, Licerio Miguel², Liliane Silvestre¹, Nilton Onari³, Ana Luiza Peloso Araujo Faim³

¹ Department of Obstetrics and Gynecology – Breast Disease Division, Ribeirão Preto Medical School, University of São Paulo. Brazil. ² Department of Gynecology & Obstetrics, Women's Health Reference Center of Ribeirão Preto (MATER), Ribeirão Preto Medical School, University of São Paulo. Brazil. ³ Department of Radiology, Hospital de Amor de Campo Grande, Mato Grosso do Sul. Brazil.

Objective: To establish the most reliable machine learning model to predict malignancy in BI- RADS; 4a and 4b breast lesions, and optimize the negative predictive value to minimize unnecessary biopsies. Methodology: We included clinical and ultrasonographic attributes from 1,250 breast lesions from four Institutions classified as BI-RADS; 3 , 4a, 4b, 4c, 5 and 6. We selected the most informative attributes to train the models in order to make inferences about the diagnosis of BI-RADS; 4a and 4b lesions (validation dataset). Using the best parameters and hyperparameters selected we tested the performance of nine models and 1530 ensemble models. Results: The most informative attributes were shape, margin, orientation and size of the lesions, the resistance index of the internal vessel, the age of the patient and the presence of a palpable lump. The highest mean NPV was achieved with XGBoost (93.6%).The final performance of the best ensemble model was: NPV= 96.4%, sensitivity= 81.5%, specificity= 84.1%, PPV= 46.8%, f1-score= 59.5% and the final accuracy= 83.7%. Age was the most important attribute to predict malignancy. The use of the final model associated with the patient’s age would reduce in 51% the number of biopsies in women with BI-RADS; 4a or 4b lesions. Conclusion: Machine learning can predict malignancy in BI-RADS; 4a and 4b breast lesions identified by the US, based on clinical and ultrasonographic features. Our final prediction model would be able to avoid 51% of the 4a and 4b breast biopsies, without missing any cancers. Keywords: ultrasonography, mammary; machine learning; artificial intelligence; image-guided biopsy