Comment on 'CAM-QUS guided self-tuning modular CNNs with multi-loss functions for fully automated breast lesion classification in ultrasound images'.

An analysis of the methodology used by the authors of the commented article is presented and errors related to data preparation are pointed out.