ContactCut: A Partition-Aware Contact Selection Framework for False-Merge Correction in Connectomics
Authors: Byunghyun Ko, Won-Ki Jeong, Chaoli Wang
Organizations: Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, IN, USA · Department of Computer Science and Engineering, Korea University, Seoul, Korea
Localized false-merge correction in connectomics requires converting an identified error into an executable native 3D partition. Still, neither intermediate correction predictions nor seeded graph solvers alone determine which native source-sink constraints should define that partition. We formulate this problem as a 3D-native partition-aware contact selection problem: selecting candidate contacts based on the complete partitions they induce. Our solution comprises two components: ContactCut (CC)-Auto, which uses learned single-contact selection, and ContactCut (CC)-Search, which enables bounded paired-contact search, generates compact alternatives, and applies learned panel ranking. On historical two-way false-merge corrections from MICrONS and FlyWire, CC-Auto improves failure-aware 3D partition (FA-3D Part.) mIoU by 5.9 and 3.4 points over the strongest graph-based controls and by 12.5 and 10.0 points over the best learned baselines. A MICrONS-trained ranker transfers to FlyWire with negligible degradation. CC-Search expands oracle reachability and prioritizes near-oracle alternatives. The code is available at https://github.com/bentake/ContactCut.
Figures & tables
Figure 1: Supervoxel contacts and candidate selection. (a) Supervoxels correspond to graph nodes. The voxel faces shared by two supervoxels form a contact, represented by a graph edge. (b) Candidate contacts have centroids within 500 nm of the localized error cue x0 (red). Shown candidate centroids are green. The crossed marker denotes an excluded contact.
Simple baselines
Graph-based
Learned baselines
Ours
Metric
NO-OP
Random
Geometric
NC-RW
NC-WS
PAT M→FW
PAT
CB2 M→FW
CB2
CC-Auto M→FW
CC-Auto
MICrONS eligible N=733 ; common-success N=575
Coverage (%)
100.0
100.0
100.0
99.2
99.2
–
100.0
–
78.6
–
99.2
FA-3D Part. mIoU ↑
0.358
0.546
0.566
0.598
0.588
–
0.532
–
0.391
–
0.657
3D Part. mIoU ↑
0.356
0.552
0.573
0.612
0.599
–
0.538
–
0.497
–
0.670
MIN ↑
0.000
0.440
0.467
0.440
0.415
–
0.425
–
0.303
–
0.558
Table 1: 3D-native automatic false-merge correction. Coverage and FA-3D Part. mIoU use the eligible TEST cohort. Eligibility requires a valid native two-way historical reference. The remaining metrics use the intersection of scorer-valid outputs across all applicable displayed methods. M→FW denotes zero-shot transfer. Best/second-best are bold / underlined .
Figure 2: Qualitative 3D-native correction results. Predictions are shown against the historical reference.
(a) Selector × solver
(b) Bounded search
Selector
GC
RW
WS
Condition
MICrONS
FlyWire
MICrONS ( N=727 )
H_FIXED base
0.641
0.639
Nearest
0.557
0.603
0.593
Best-single †
0.843
0.798
Auto
0.663
0.654
0.643
B45-pair †
0.866
0.837
FlyWire ( N=2906 )
Best-in- K5†
0.826
0.800
Nearest
0.562
0.612
0.612
Table 2: Contact-selection ablation and bounded search. 3D Part. mIoU on contact-valid TEST operations; † denotes reference-best reachable or best-in-panel quality.
Figure 3: Effect of contact selection. The red marker in (a) is the localized cue x0 . This example shows how nearby contacts induce different partitions under the same graph-cut setup. e2 is selected by CC-Auto.
Proofreading--correcting segmentation errors in 3D brain reconstructions--is the rate-limiting step in synapse-resolution connectomics. We release ConnectomeBench2, a unified multi-species dataset of over 716,485 expert-labeled proofreading decisions with >4,500,000 associated images spanning four major open connectomes (mouse, human, zebrafish, fly), spanning both split and merge error correction. Trained on this dataset, a single Vision Transformer with shared encoders for mesh geometry and electron microscopy reaches human-level accuracy across species for split error correction and merge error identification, with performance scaling with data size and modality. Beyond accuracy, we show that the model is well-calibrated within distribution, that measures of distribution distance predict where calibration and accuracy will degrade on unseen data, and that connectomics-specific pretraining and active learning-based sample selection show potential to substantially reduce the labeling effort needed to extend to new species and brain regions. The benchmark provides the infrastructure to train and evaluate increasingly capable vision models for connectomic proofreading. Data and code availability. The ConnectomeBench2 dataset is released on Hugging Face at https://huggingface.co/datasets/jeffbbrown2/ConnectomeBench2. The accompanying codebase is available on GitHub at https://github.com/timfarkas/ConnectomeBench2.
Locating genomic features from genomic contact maps, such as centromere identification from genome-wide chromosome conformation capture techniques, notably Hi-C, can be formulated as an inverse problem: infer one parameter per entity given a map summarizing pairwise interactions through blocks of variable numbers and sizes. In this work, we introduce a data-driven approach that leverages shared structure between these contact maps, such as global alignment between localized patterns, while handling the variability in number and size of chromosomes arising in real-world data. Our approach relies on a transformer architecture capable of handling such variability and a custom simulator to generate abundant, yet computationally cheap synthetic data for training. Applied to the problem of centromere localization, the method recovers genomic positions at or below the map resolution across species with various genome sizes using a coarse-to-fine refinement step when chromosome sizes fall outside the training range. We concentrate our evaluation on Hi-C data, where the block structure is well characterized and ground truth is available.
Eloïse Touron, Pedro L. C. Rodrigues, Julyan Arbel +2
Univ. Grenoble Alpes, Inria CNRS, Grenoble INP, LJK, France · TIMC, Univ. Grenoble Alpes CNRS, Grenoble INP, France
We present Linkify, a framework for learning from interface-augmented assembly graphs to enable context-aware part retrieval in mechanical assemblies. While recent generative AI methods for CAD have focused largely on isolated parts or monolithic assemblies, the rich geometric information at the interfaces between parts, where function is realized, remains underexplored. We address this gap by recomputing high-fidelity interface geometry for the Fusion 360 Gallery Assembly dataset, correcting missing and erroneous contacts, and generating point-cloud representations of local contact regions. Using this data, we construct assembly graphs whose nodes encode part geometry and whose edges encode interface geometry via a pretrained point-cloud encoder. On top of this representation, we train a Graph Attention Network based on GATv2 to solve a masked part prediction task: given an assembly with one part held out, the model predicts the class of the missing component from a large vocabulary of geometrically clustered parts, thereby approximating a realistic part-retrieval scenario. Compared to non-graph baselines such as logistic regression and k-nearest neighbors operating on aggregated node features, Linkify achieves higher Top-K accuracy and F1 scores. Ablation studies on graph connectivity, edge attributes, and attention mechanisms demonstrate that accurate contact computation and dynamic attention over interfaces are critical for performance. Our corrected interface dataset and training pipeline, released publicly, provide a foundation for future interface-aware models for assembly retrieval, validation, and generative design.
Anushrut Jignasu, Daniele Grandi
Iowa State University, USA · Autodesk Research, USA