유방암 분류를 위한 합성곱 신경망 및 랜덤 포레스트 분류기의 비교 연구

다국어 초록 (Multilingual Abstract)

This study compared the performance of convolution neural networks and random forest models for breast cancer diagnosis using breast ultrasound images. CNN has smooth learning of complex patterns with high accuracy and sensitivity, and RF stands out in terms of interpretability and computational efficiency. The performance of each model was evaluated through various indicators (accuracy, sensitivity, specificity, AUC, F1 Score, etc.). CNN derived favorable results in interpretability for each of the RF in complex image pattern learning. Based on the strengths and limitations of the two models, a plan to improve the automatic breast cancer diagnosis system was suggested, and a plan to improve model performance through data augmentation and trasfer learning was discussed.