Projects & Workflows
Research themes pursued by CTTIR. Each project card brings together the relevant Software packages and the Workflows below that demonstrate them in practice — the full path from method development to applied analysis on a defined research question.
Software cards live at Software.
Projects
Hyperspectral Imaging
Label-free tissue characterization from spectral signatures. The hyperloom suite brings together CTTIR's hyperspectral packages: camera-specific readers for Cubert and TIVITA data, a common spectral cube with preprocessing and analysis, and spectral visualisation.
hyperloom— suite package for CTTIR hyperspectral imagingcuvis.r— reading Cubert hyperspectral measurementstivis.r— reading TIVITA hyperspectral recordingshyperspectR— spectral cubes, preprocessing and analysishyperspectaculR— spectral visualisation
Spatial Biology
Cells in their tissue context. The phenoloom suite brings together CTTIR's spatial biology packages: cell segmentation, single-cell profiles, gating of cell populations and spatial neighbourhood analysis.
phenoloom— suite package for CTTIR spatial biologysegmantR— cell segmentation and mask correctioncellspecR— single-cell profilesgatekeepR— gating cell populationsphenoscapR— spatial neighbourhoods
Workflows
Topic-driven worked examples showing the CTTIR R packages in real use — each workflow tackles a concrete biomedical question end-to-end, with multiple CTTIR packages composed into a single reproducible pipeline.
Each workflow is a standalone workflowr project — fully versioned and reproducible, rendered as a static site with code, results, and narrative side by side. Curated demonstrations, not exhaustive documentation. Clone, adapt, extend.
Hyperspectral Tissue Characterization — From Spectral Cube to Tissue Class Map
WalkthroughCan tissue regions be told apart by their reflectance spectra alone, and which spectral features separate them?
Executed walkthrough of the hyperloom suite: read a TIVITA recording with tivis.r and check the Cubert SDK through cuvis.r, then preprocess, fit hemoglobin, embed and classify simulated tissue scenes with hyperspectR, holding out whole scenes, and summarise the cube with hyperspectaculR. The data are synthetic with a known answer, so every step can be checked.
Featured packages:
hyperloom, tivis.r, cuvis.r, hyperspectR, hyperspectaculR
What you'll learn
- Reading TIVITA recordings and metadata into an hsi_cube with tivis.r
- Reusable preprocessing recipes (Savitzky–Golay and SNV) with hyperspectR
- Hemoglobin fitting and band ratios, checked against known simulated values
- PCA embeddings whose loadings read as spectra, and unsupervised region recovery
- SVM classification evaluated with whole scenes held out
Spatial Tissue Phenotyping — From Images to Cell Neighbourhoods
WalkthroughWhich cell populations are present in a tissue section, and how are they arranged relative to each other?
Executed walkthrough of the phenoloom suite on a synthetic seven-channel immunofluorescence section: segmantR segments nuclei and cells, cellspecR turns the measurements into a validated, checksummed cell table, gatekeepR estimates marker cut-offs and assigns cell types, and phenoscapR maps neighbour enrichment, distances and cellular neighbourhoods. Each step is checked against the drawn ground truth.
Featured packages:
phenoloom, segmantR, cellspecR, gatekeepR, phenoscapR
What you'll learn
- Segmenting nuclei by watershed and growing them into cells with segmantR, scored against known nuclei
- Reading segmantR features into a validated, checksummed cell table with cellspecR
- Declaring a marker panel and phenotype hierarchy in gatekeepR, with an unstained marker reported as not callable
- Testing which cell types sit together and finding cellular neighbourhoods with phenoscapR
- Checking every step against the ground truth of a synthetic image