Researchers from China evaluated the use of artificial intelligence (AI) algorithms for different breast ultrasound modalities, finding that AI in combination with 3D ultrasound and color-flow Doppler was most accurate in diagnosing breast cancer, according to a new study published in the Chinese Medical Journal.
Researchers collected 15,648 images from 3,623 patients and used half of them to train three different convolutional neural network (CNN) models, while the other half were used to test the algorithms. They tested the use of AI on three types of ultrasound acquisitions:
- 2D ultrasound images
- 2D ultrasound plus color-flow Doppler; color-flow Doppler provided information on blood flow surrounding lesions
- 2D ultrasound plus color-flow Doppler plus pulsed-wave Doppler; pulsed-wave Doppler provided spectral information over a specific area within the lesions
The team found that the accuracies for all three were similar, but the second model, with 2D images and color-flow Doppler, performed slightly better than the other two, with an accuracy of around 88%.
The performance of the CNN models was compared with the interpretations of 37 radiologists on a subset of 50 randomly selected images. The CNN model had an accuracy of 89.2%, with a processing time of less than two seconds. In contrast, the average accuracy of the ultrasonologists was 30%, with an average time of 314 seconds.
