Trustworthy clinical AI solutions: A unified review of uncertainty quantification in Deep Learning models for medical image analysis.

The full acceptance of Deep Learning (DL) models in the clinical field is rather low with respect to the quantity of high-performing solutions reported in the literature. End users are particularly reluctant to rely on the opaque predictions of DL models. Uncertainty quantification methods have been proposed in the literature as a potential solution, to reduce the black-box effect of DL models and increase the interpretability and the acceptability of the result by the final user. In this review, we propose an overview of the existing methods to quantify uncertainty associated with DL predictions. We focus on applications to medical image analysis, which present specific challenges due to the high dimensionality of images and their variable quality, as well as constraints associated with real-world clinical routine. Moreover, we discuss the concept of structural uncertainty, a corpus of methods to facilitate the alignment of segmentation uncertainty estimates with clinical attention. We then discuss the evaluation protocols to validate the relevance of uncertainty estimates. Finally, we highlight the open challenges for uncertainty quantification in the medical field.

THAP11F80L cobalamin disorder-associated mutation reveals normal and pathogenic THAP11 functions in gene expression and cell proliferation.

Twelve human THAP proteins share the THAP domain, an evolutionary conserved zinc-finger DNA-binding domain. Studies of different THAP proteins have indicated roles in gene transcription, cell proliferation and development. We have analyzed this protein family, focusing on THAP7 and THAP11. We show that human THAP proteins possess differing homo- and heterodimer formation properties and interaction abilities with the transcriptional co-regulator HCF-1. HEK-293 cells lacking THAP7 were viable but proliferated more slowly. In contrast, HEK-293 cells were very sensitive to THAP11 alteration. Nevertheless, HEK-293 cells bearing a THAP11 mutation identified in a patient suffering from cobalamin disorder (THAP11F80L) were viable although proliferated more slowly. Cobalamin disorder is an inborn vitamin deficiency characterized by neurodevelopmental abnormalities, most often owing to biallelic mutations in the MMACHC gene, whose gene product MMACHC is a key enzyme in the cobalamin (vitamin B12) metabolic pathway. We show that THAP11F80L selectively affected promoter binding by THAP11, having more deleterious effects on a subset of THAP11 targets, and resulting in altered patterns of gene expression. In particular, THAP11F80L exhibited a strong effect on association with the MMACHC promoter and led to a decrease in MMACHC gene transcription, suggesting that the THAP11F80L mutation is directly responsible for the observed cobalamin disorder.