A study using routine breast ultrasound images trained a deep learning radiomics model to predict sentinel lymph node metastasis preoperatively. Results showed moderate discrimination across cohorts, while the authors emphasized the need for larger prospective validation before clinical use. The work reflects a continued shift toward computer-aided triage in imaging, aiming to reduce unnecessary procedures and tighten surgical planning. For the industry, it’s a reminder that performance alone is not enough—cohort expansion and workflow integration are the next hurdles.