The study presents a new system for semi-automated breast ultrasound report generation that combines artificial intelligence and deep learning to reduce the workload of radiologists. The research was carried out on a sample of 2,119 elastographic images and 60 annotated patient cases from two American ultrasound machines. The system achieved high accuracy in elastography classification with values under the receiver operating characteristic curve of 0.92, 0.91, and 0.88 for different display types. When tested, the reporting module correctly identified all suspicious masses on both instruments with 100% sensitivity in lesion detection. The average report generation time was 31 seconds per patient for the GE Healthcare device and 36 seconds for the Supersonic Image device. The proposed framework enables accurate and efficient generation of reports adapted to different devices and demonstrates the potential for deployment in clinical practice.