Hybrid deep feature and machine learning framework for classification of thyroid nodules in ultrasound images

1

The study proposes a hybrid computer-aided diagnosis framework for classifying thyroid nodules using deep learning and machine learning techniques.
2

The framework combines deep transfer learning with gradient-boosted decision tree classification to enhance diagnostic accuracy.
3

Experimental results show that the proposed method outperforms conventional approaches in distinguishing benign from malignant thyroid nodules.
4

The framework demonstrates robustness across variations in ultrasound appearance without extensive parameter tuning.
5

The findings suggest the method's potential for integration into clinical workflows to improve decision-making and reduce unnecessary interventions.

1