Data associated with NucleiSky enables cross-scale multimodal registration of microscopy data using nuclei constellations

This Zenodo archive contains raw microscopy images, segmentation masks, representative query images or crops, channel-transfer outputs, and benchmarking results associated with the manuscript: NucleiSky enables cross-scale multimodal registration of microscopy data using nuclei constellations NucleiSky is a microscopy image-registration framework that uses the spatial arrangement of nuclei, or other segmented landmarks, as a coordinate system to localise query fields of view within larger reference images. The datasets deposited here support the benchmarking, validation, and example applications presented in the manuscript, including 2D localisation benchmarks, cross-scale fluorescence registration, live-to-fixed registration using synthetic nuclear labels, and 3D volumetric localisation. Each dataset is provided as an individual ZIP archive. All ZIP files are deposited together within the same Zenodo record. Each ZIP archive contains a dataset-specific README.md file describing the files, acquisition context, segmentation strategy, and how the dataset was used in the manuscript. Archive organisation The Zenodo record contains the following dataset ZIP files: Dataset 1.zip Dataset 2.zip Dataset 4.zip Dataset 5.zip Dataset 6.zip Dataset 7.zip Dataset 8.zip Dataset 10.zip Datasets 3 and 9 are based on publicly available external resources cited in the manuscript. Dataset 1: 2D localisation benchmark, DAPI-stained HUVEC monolayer This ZIP contain the raw data and outputs for Dataset 1, a tiled widefield fluorescence image of a fixed DAPI-stained HUVEC monolayer. This dataset was used in the NucleiSky 2D localisation benchmark (Figure 2A) and the segmentation-error robustness analysis (Figure 2E). The ZIP should contain: README.md Image1.tif: full reference image. masks_full_original_scale.tif: segmentation mask of the full reference image. random_crop.tif: example query crop extracted from the reference image. masks_crop_original_scale.tif: segmentation mask corresponding to the example crop. benchmarking/: benchmarking results for Dataset 1 shown in Figure 2. benchmark segmentation/: segmentation-error robustness benchmarking results shown in Figure 2E. Dataset 2: 2D localisation benchmark, DAPI-stained HUVEC/SU.86.86 montage This ZIP contain the raw data and outputs for Dataset 2, a spinning-disk confocal montage of a fixed DAPI-stained HUVEC monolayer co-incubated with SU.86.86 pancreatic cancer cells. This dataset was used in the NucleiSky 2D localisation benchmark shown in Figure 2B. The ZIP contains: README.md Image2.tif: full reference image. masks_full_original_scale.tif: segmentation mask of the full reference image. random_crop.tif: example query crop extracted from the reference image. masks_crop_original_scale.tif: segmentation mask corresponding to the example crop. benchmarking/: benchmarking results for Dataset 2 shown in Figure 2. Dataset 4: 2D localisation benchmark, DAPI-stained Escherichia coli This ZIP contains the raw data and outputs for Dataset 4, a tiled widefield fluorescence image of fixed DAPI-stained Escherichia coli. This dataset was used in Figure 2D to test whether the NucleiSky constellation-matching principle could be extended beyond mammalian nuclei, using bacterial objects as spatial landmarks. The ZIP contains: README.md ecoli_1.tif: full reference image. ecoli_1_mask_full.tif: segmentation mask of the full reference image. benchmarking/: benchmarking results for Dataset 4 shown in Figure 2. Dataset 5: Cross-scale localisation, 63× query into 20× overview This ZIP contains the raw data and outputs for Dataset 5, a cross-scale localisation example in which a high-magnification 63× fluorescence query image was localised within a lower-magnification 20× overview mosaic of a DCIS.com epithelial monolayer. This dataset was used in Figure 3A-B. The ZIP contains: README.md 2024-03-20_Box3_20x-Airyscan Processing-01-1.tif: 20× reference overview image. MAX_2024-03-20_Box3_63x-Airyscan Processing-03-1-1.tif: high-resolution 63× query image, generated as a maximum-intensity projection. 2024-03-20_Box3_63x-Airyscan Processing-03.tif: multichannel high-resolution image used for channel transfer after the NucleiSky transformation was found. Dataset 6: Cross-scale localisation, 63× query into 10× overview This ZIP contains the raw data and outputs for Dataset 6, a cross-scale localisation example in which a high-magnification 63× fluorescence query image was localised within a lower-magnification 10× overview mosaic of a DCIS.com epithelial monolayer. This dataset was used in Figure 3C-D. The ZIP contains: README.md 2024-03-19_Box1_10x-Airyscan Processing-01-1.tif: 10× reference overview image. 2024-03-19_Box1_63x-Airyscan Processing-04-1.tif: high-resolution 63× query image. 2024-03-19_Box1_63x-Airyscan Processing-04-1stack.tifA high-resolution multichannel stack used for channel transfer after the NucleiSky transformation was found. Dataset 7: Live-to-fixed registration using synthetic nuclear labels This ZIP contains the raw data and outputs for Dataset 7, a live-to-fixed registration example in which a label-free bright-field movie was registered to a post-fixation fluorescence tile scan using a pix2pix-derived synthetic nuclear image. This dataset was used in Figure 4. The ZIP contains: README.md 15001 - HPMECs_AsPC-1-0..nd2 (series 1)_last frame.tif: last frame of the live bright-field video. 20260213_HPMECs_AsPC-1_prediction.tif: pix2pix-derived DAPI-like synthetic nuclear image used as the NucleiSky query image. HPMECs_AsPC-1-0_Frames_14702-15001.tif: final 300 frames of the live bright-field video. MAX_20260223_HPMEC_AsPC-1_MTG_Nuclei.tif: fixed-cell Hoechst nuclear reference image. MAX_20260223_HPMEC_AsPC-1_MTG_16-bit.tif: multichannel fixed-cell fluorescence image used for channel transfer after the NucleiSky transformation was found. Dataset 10: 3D NucleiSky benchmark, DAPI-stained mouse brain slice This ZIP contains the raw data and segmentation labels for Dataset 10, a DAPI-stained mouse brain slice acquired as a spinning-disk confocal Z-stack. This dataset was used in the NucleiSky3D volumetric localisation benchmark shown in Figure 6E. The ZIP contains: README.md 221116_Brain_TLNRD1_CCM3.sld - Brain_PECAM_CCM3(2)-1.tif: 3D image dataset containing the DAPI channel used to generate the nuclear landmark reference. labels_full.tif: segmentation labels for the full 3D image dataset.