Detecting and segmenting nuclei in fluorescence microscopy images is a very common and important workflow, allowing for counting nuclei, measuring cell density in a tissue, and quantifying nuclear morphology. This repository is a demonstration of a nuclear segmentation workflow using pre-trained artificial intelligence models, conversion to data tabular data, and subsequent data analysis.
The raw data consists of 200 16-bit TIFF images from U2OS cells, each with a field of view containing numerous nuclei. Each image represents the DNA channel and shows Hoechst-stained nuclei. Six sample images from the dataset are shown below. Each bright blob is a single cell nucleus.
While digital image processing using common filters, edge detectors, and other classic algorithms can be used for segmentation, one of the StarDist pre-trained models is able to very quickly and accurately produce nuclei labels. As seen in the two example pairs below (left - raw data; right - AI prediction), the segmentation accuracy is very high.
Now that the nuclei have been segmented, the label images can be read and constructed into a tabular format, in this case, a Polars DataFrame. The first 10 lines of this data can be seen in the table below, including columns for area, aspect ratio, perimeter and solidity, for each individual detected cell nucleus. It is important to note that for each label image, nuclei touching the image border were excluded because these would adversely affect the aforementioned morphological measurements.
┌─────────────────────────────────┬────────────────┬───────┬────────┬──────────────┬────────────┬──────────┐
│ filename ┆ img_total_area ┆ label ┆ area ┆ aspect_ratio ┆ perimeter ┆ solidity │
│ --- ┆ --- ┆ --- ┆ --- ┆ --- ┆ --- ┆ --- │
│ str ┆ i64 ┆ i64 ┆ f64 ┆ f64 ┆ f64 ┆ f64 │
╞═════════════════════════════════╪════════════════╪═══════╪════════╪══════════════╪════════════╪══════════╡
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 1 ┆ 835.0 ┆ 0.631725 ┆ 110.325902 ┆ 0.97093 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 2 ┆ 1495.0 ┆ 0.684223 ┆ 149.296465 ┆ 0.96952 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 3 ┆ 756.0 ┆ 0.829363 ┆ 100.911688 ┆ 0.978008 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 4 ┆ 688.0 ┆ 0.74359 ┆ 97.740115 ┆ 0.967651 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 5 ┆ 1090.0 ┆ 0.922205 ┆ 121.63961 ┆ 0.976703 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 6 ┆ 760.0 ┆ 0.905834 ┆ 101.740115 ┆ 0.970626 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 7 ┆ 1081.0 ┆ 0.774601 ┆ 122.953319 ┆ 0.972997 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 8 ┆ 1370.0 ┆ 0.811672 ┆ 138.710678 ┆ 0.968883 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 10 ┆ 803.0 ┆ 0.675087 ┆ 108.083261 ┆ 0.961677 │
│ IXMtest_A02_s1_w1051DAA7C-7042… ┆ 361920 ┆ 11 ┆ 834.0 ┆ 0.55059 ┆ 113.497475 ┆ 0.964162 │
└─────────────────────────────────┴────────────────┴───────┴────────┴──────────────┴────────────┴──────────┘
The data distributions for nuclear area, aspect ratio, and solidity are shown below. This dataset contains no experimental groupings.
