Cancer research and therapeutics is increasingly being approached from an immunological perspective and the number of samples in research and diagnostics is growing.
This coupled with the issues of bias and subjectivity in image analysis and the demand for more quantitative information are calling for a need to transform how we look at cancer — a need to automate, standardize, and optimize.
How does Aiforia enable cancer research?
Aiforia’s deep learning AI platform has successfully enabled the automation of a versatile range of analyses in a number of areas and sample types such as in immuno-oncology, breast, prostate, lung, and testis.
With Aiforia you can automate the detection of relevant regions of interest based on morphology, use external ground-truths (e.g. treated vs non-treated groups) to find these regions in precise locations, and visualize the feature of interest.
Cancer biomarkers PD-L1, ER, Ki67 and HER2
Tumor-infiltrating lymphocyte (TIL) percentage
Of spatial and morphological metrics
Cell-to-tumor distance, for example: between stromal CD8+ cells and lung cancer tumor borders (see Figure 1)
Cell-to-cell distances, including unique cell pairs (only the closest cells), for example: PD1 and PD-L1 pairing in lung stroma (see Figure 2)
Precise and quantitative grading of whole slide images
Tumor grading and tumor burden
Regression models predicting continuous values
Cell and nucleus size, patient survival time and other end-points