The artificial intelligence automatic diagnosis model based on the needle tissue of prostate ultrasound

Aiming at the problems of high noise interference, low signal-to-noise ratio and low resolution often faced by pathological classification of prostate ultrasound images, a complete dataset of prostate ultrasound image needle tract tissues available for AI was constructed by taking prostate ultrasound image puncture needle tract tissues as an example. The Mask R-CNN is used to achieve the segmentation of prostate ultrasound image puncture needle tract tissue, and the established needle tract segmentation model can automatically identify the parts of prostate tissue that may be cancer. Then, the background information is masked and the prostate ultrasound image needle tract tissue dataset is constructed for use in each classical image classification detection network model. Comparing the four mainstream comparative learning methods, the performance metrics such as Precision, Recall, and F1-score are all higher than directly using the original image classification. The constructed image needle channel dataset has higher accuracy in each classical image classification detection can effectively improve the efficiency of prostate cancer diagnosis work, and promote the development of prostate cancer ultrasound image diagnosis technology.