The Flow has not summarised this story yet — read it at arXiv Machine Learning.
arXiv:2604. 27277v3 Announce Type: replace-cross Abstract: Brain MRI underpins a wide range of neuroscientific and clinical applications, yet most learning-based methods remain task-specific and require substantial labeled data.
arXiv:2607. 11962v1 Announce Type: cross Abstract: Self-supervised learning offers a compelling approach for medical imaging, where labeled data are scarce and acquisition costs are high.
arXiv:2606. 05870v1 Announce Type: cross Abstract: Neurodegenerative disorders such as Alzheimer's disease exhibit highly organized patterns of regional brain vulnerability, yet the biological mechanisms underlying this spatial selectivity remain incompletely understood.
The paper introduces SYMBA, a self‑supervised pretraining framework tailored for structural T1‑weighted brain MRI. SYMBA uses anatomically constrained masking, a multi‑scale structural similarity reconstruction loss, and a progressive curriculum to capture distributed morphometric patterns. Evaluated on nine heterogeneous cohorts and two unseen datasets, SYMBA outperforms existing neuroimaging foundation models, achieving high AUROC scores for Alzheimer’s disease classification, mild cognitive impairment conversion, and amyloid status prediction, and shows strong sample efficiency in low‑label settings.
FiberGeoText (FGT) is a vision‑language model that clusters short‑range superficial white matter streamlines from ultra‑high‑resolution diffusion MRI into population‑level groups. It jointly encodes each streamline’s 3‑D trajectory, cortical anatomical context (via text from multiple parcellation schemes), and shape, using a pretrained large language model to unify heterogeneous anatomical descriptions. Evaluations on 0.76 mm diffusion data show that FGT outperforms state‑of‑the‑art methods in cortical parcel coherence, shape consistency, cluster‑size consistency, and cross‑subject correspondence, and it generalizes well to unseen subjects, recovering 96.7 % of learned clusters.
The paper investigates how ESMFold’s folding trunk transforms amino‑acid sequences into 3D protein structures by using sparse autoencoders (SAEs) to uncover interpretable features. It first analyzes representation distributions across modules and recycles to identify distinct spatial computations, then shows that single representations carry inherited semantics while pair representations compute pairwise geometry. Finally, the authors use a reasoning language model and gradient‑based attribution to trace concept composition across blocks, verifying alignment with known biochemistry and demonstrating that the trained sparse dictionaries can predict the model’s fold quality, thereby turning the black‑box trunk into an interpretable, scientifically verifiable circuit.