Mycosis fungoides is characterized by widespread skin patches that may progress to plaques and tumors, necessitating precise tumor burden assessment for staging and treatment guidance. However, existing methods, including the widely accepted modified Severity Weighted Assessment Tool (mSWAT), present significant challenges in routine practice owing to their time-consuming nature and interobserver variability. This study developed an artificial intelligence model, mSWAT-Net, to estimate mSWAT scores using clinical images of patients with mycosis fungoides. Notably, the overlap area segmentation submodule of mSWAT-Net addressed double-counting errors in multiangle photos through training on 3904 annotated images generated from 61 three-dimensional human images. Across 2463 standardized full-body photographs from 134 imaging series, mSWAT-Net demonstrated performance comparable with that of experienced cutaneous lymphoma specialists, achieving intraclass correlation coefficients of 0.917 (internal validation) and 0.846 (temporal validation) for mSWAT score. Moreover, mSWAT-Net outperformed 3 junior dermatologists in image-based scoring (intraclass correlation coefficient = 0.917 vs 0.777) and demonstrated robust performance when compared with ground truth derived from 3-dimensional patient imaging (intraclass correlation coefficient = 0.812). Finally, mSWAT-Net was deployed as a free web application to support mycosis fungoides management in clinical settings. These findings highlight the potential of mSWAT-Net as an accurate, automated clinical tool for facilitating patient follow-up, treatment monitoring, and remote consultations.
Trace elements and isotopes (TEIs) are important to marine life and are essential tools for studying ocean processes1. Two different frameworks have arisen regarding marine TEI cycling: reversible scavenging favours water-column control on TEI distributions2–5, and seafloor boundary exchange emphasizes sedimentary imprints on water-column biogeochemistry6,7. These two views lead to disparate interpretations of TEI behaviours8–10. Here we use rare earth elements and neodymium isotopes as exemplar tracers of particle scavenging11 and boundary exchange6,7,12. We integrate these data with models of particle cycling and sediment diagenesis to propose a general framework for marine TEI cycling. We show that, for elements with greater affinity for manganese oxide than biogenic particles, scavenging is a net sink throughout the water column, contrary to a common assumption for reversible scavenging3,13. In this case, a benthic flux supports increasing elemental concentrations with water depth. This sedimentary source consists of two components: one recycled from elements scavenged by water-column particles, and another newly introduced to the water column through marine silicate weathering inside sediment8,14,15. Abyssal oxic diagenesis drives this benthic source, and exerts a strong influence on water-column biogeochemistry through seafloor geometry and bottom-intensified turbulent mixing16,17. Our findings affirm the role of authigenic minerals, often overshadowed by biogenic particles, in water-column cycling18, and suggest that the abyssal seafloor, often regarded as inactive, is a focus of biogeochemical transformation19,20.
In this paper, we develop a new adaptive hyperbolic-cross-space mapped Jacobi (AHMJ) method for solving multidimensional spatiotemporal integrodifferential equations in unbounded domains. By devising adaptive techniques for sparse mapped Jacobi spectral expansions defined in a hyperbolic cross space, our proposed AHMJ method can efficiently solve various spatiotemporal integrodifferential equations such as the anomalous diffusion model with reduced numbers of basis functions. Our analysis of the AHMJ method gives a uniform upper error bound for solving a class of spatiotemporal integrodifferential equations, leading to effective error control.