Knowledge-driven AI-generated data for accurate and interpretable breast ultrasound diagnoses
This work addresses a critical bottleneck in medical imaging for radiologists by improving diagnostic accuracy in rare cases, though it is incremental as it builds on existing generative methods.
The study tackled the problem of inaccurate diagnoses for rare cases in breast ultrasound due to long-tailed data distributions by introducing TAILOR, a pipeline that generates synthetic data for rare cases, resulting in a diagnostic model that outperformed nine radiologists by 33.5% in specificity with the same sensitivity and significantly improved performance on ductal carcinoma in situ.
Data-driven deep learning models have shown great capabilities to assist radiologists in breast ultrasound (US) diagnoses. However, their effectiveness is limited by the long-tail distribution of training data, which leads to inaccuracies in rare cases. In this study, we address a long-standing challenge of improving the diagnostic model performance on rare cases using long-tailed data. Specifically, we introduce a pipeline, TAILOR, that builds a knowledge-driven generative model to produce tailored synthetic data. The generative model, using 3,749 lesions as source data, can generate millions of breast-US images, especially for error-prone rare cases. The generated data can be further used to build a diagnostic model for accurate and interpretable diagnoses. In the prospective external evaluation, our diagnostic model outperforms the average performance of nine radiologists by 33.5% in specificity with the same sensitivity, improving their performance by providing predictions with an interpretable decision-making process. Moreover, on ductal carcinoma in situ (DCIS), our diagnostic model outperforms all radiologists by a large margin, with only 34 DCIS lesions in the source data. We believe that TAILOR can potentially be extended to various diseases and imaging modalities.