Detecting out-of-distribution (OOD) samples for trusted medical image segmentation remains a significant challenge. The critical issue here is the lack of a strict definition of abnormal data, which often results in artificial problem settings without measurable clinical impact. In this paper, we redesign the OOD detection problem according to the specifics of volumetric medical imaging and related downstream tasks (e.g., segmentation). We propose using the downstream model's performance as a pseudometric between images to define abnormal samples. This approach enables us to weigh different samples based on their performance impact without an explicit ID/OOD distinction. We incorporate this weighting in a new metric called Expected Performance Drop (EPD). EPD is our core contribution to the new problem design, allowing us to rank methods based on their clinical impact. We demonstrate the effectiveness of EPD-based evaluation in 11 CT and MRI OOD detection challenges.

We consider the problem of inductive matrix completion under the assumption that many features are non-informative, which leads to row- and column-sparse structure of coefficient matrix. Under the additional assumption on the low rank of coefficient matrix we propose the matrix factorization framework with group-lasso regularization on parameter matrices. We suggest efficient optimization algorithm for the solution of the obtained problem. From theoretical point of view, we prove the oracle generalization bound on the expected error of matrix completion. Corresponding sample complexity bounds show the benefits of the proposed approach compared to competitors in the sparse problems. The experiments on synthetic and real-world datasets show the state-of-the-art efficiency of the proposed method.