Research map / Image generation / Autoregressive and latent generation
Toward Generalizable Whole Brain Representations with High-Resolution Light-Sheet Data
Minyoung E. Kim, Dae Hee Yun, Aditi V. Patel, Madeline Hon, Webster Guan, Taegeon Lee, Brian Nguyen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train detector on a different marker than the test marker | ConvMixer cell detector · CANVAS cross-marker evaluation | F1: TH model peaked at 0.31 F1 on cFos vs 0.83 in region 6 of TH | Hurt |
| use NeuN-trained detector on the PV dataset instead of the PV model | ConvMixer cell detector · CANVAS PV | F1: 0.63 -> 0.89 | Improved |
| use Iba1-trained detector on GFAP instead of the GFAP model | ConvMixer cell detector · CANVAS GFAP | F1: 0.33 -> 0.43 | Improved |
| lower mask ratio (0.15-0.55 instead of 0.75) (ablation) | 3D-MAE · CANVAS 3D-MAE | reconstruction loss | Improved |
| use smallest crop/patch config 16x32x32 / 4x8x8 (ablation) | 3D-MAE · CANVAS 3D-MAE | reconstruction loss | Improved |
| use 24x48x48 / 6x12x12 crop/patch for GFAP (ablation) | 3D-MAE · CANVAS GFAP | reconstruction loss | Mixed |
| train one all-markers model on 10x more data instead of marker-specific models (ablation) | 3D-MAE · CANVAS 3D-MAE | reconstruction loss: 0.0061 -> 0.0070 | No clear effect |
| add 3D-MAE features to detector features for TP/FP classification (ablation) | ConvMixer GFAP detector · CANVAS GFAP | mean F1: +22.9% mean F1, up to +86.3% on region 3 | Improved |
| add 3D-MAE features to detector features for TP/FP classification (ablation) | ConvMixer NeuN detector · CANVAS NeuN | F1: +3.0% in region 2 (r = 0.3) | Mixed |
| add content-aware variance weighting to MAE loss | 3D-MAE · CANVAS 3D-MAE | reconstruction loss | No clear effect |
Closest papers by meaning
- Micro-Bench: A Microscopy Benchmark for Vision-Language Understanding (NeurIPS 2024 · similarity 0.86)
- MorphoHELM: A Comprehensive Benchmark for Evaluating Representations for Microscopy-Based Morphology Assays (NeurIPS 2026 · similarity 0.85)
- Retrospective for the Dynamic Sensorium Competition for predicting large-scale mouse primary visual cortex activity from videos (NeurIPS 2024 · similarity 0.85)
- ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy (ICML 2025 · similarity 0.84)
- Towards Scalable Context-Aware Single-Cell Spatial Transcriptomics Prediction from Histology Images (NeurIPS 2026 · similarity 0.84)
- FISBe: A Real-World Benchmark Dataset for Instance Segmentation of Long-Range Thin Filamentous Structures (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
