Pre and post-hoc diagnosis and interpretation of malignancy from breast DCE-MRI

•We introduce a new nomenclature to categorise detection and classification methods: pre-hoc and post-hoc.•We introduce a new post-hoc method based on meta-training guided by curriculum learning, and a 1-class saliency detector.•We introduce a new pre-hoc approach that uses deep reinforcement learning for detecting lesions.•We also introduce a thorough comparison between the pre- and post-hoc approaches for breast screening from breast DCE-MRI.•Results show that post-hoc methods produce more accurate diagnosis, while pre-hoc methods are better at detecting the lesions. [Display omitted] We propose a new method for breast cancer screening from DCE-MRI based on a post-hoc approach that is trained using weakly annotated data (i.e., labels are available only at the image level without any lesion delineation). Our proposed post-hoc method automatically diagnosis the whole volume and, for positive cases, it localizes the malignant lesions that led to such diagnosis. Conversely, traditional approaches follow a pre-hoc approach that initially localises suspicious areas that are subsequently classified to establish the breast malignancy – this approach is trained using strongly annotated data (i.e., it needs a delineation and classification of all lesions in an image). We also aim to establish the advantages and disadvantages of both approaches when applied to breast screening from DCE-MRI. Relying on experiments on a breast DCE-MRI dataset that contains scans of 117 patients, our results show that the post-hoc method is more accurate for diagnosing the whole volume per patient, achieving an AUC of 0.91, while the pre-hoc method achieves an AUC of 0.81. However, the performance for localising the malignant lesions remains challenging for the post-hoc method due to the weakly labelled dataset employed during training.