Background

In traditional lung cancer diagnosis, a sample of living tissue is extracted from the patient through lung biopsy.

The sample is prepared into a Whole Slide Image (WSI) and is examined by an expert pathologist under a microscope. The report provides information to determine which treatment to undertake.

We have developed the necessary data pipeline and deep learning model for this process. The following picture shows one of the biopsy slide.

Model Prediction

Example 1



Example 2

MAKE A DIFFERENCE

We use state-of-the-art deep learning architecture developed by Google.

Using data from International Conference on Image Analysis and Recognition 2018 (ICIAR 2018). We achieve 0.95 F1-score.

The model helps pathologist in screening whole slide images in cell level and detect areas with anomalous cells.

SAVE TIME

[expand]

​SCANCER can run 24/7 to spot cancer cells.

Compared with the traditional process, time and cost can be minimised.

[/expand]

SAVE EFFORT

[expand]

Pathologists can better utilize their time to screen as much as cases and work on researches time as fast as they can.

[/expand]

SAVE KNOWLEDGE

[expand]

Preserve pathologist knowledge across different countries and generation.

Our vision is to accumulate expert experience and build a worldwide knowledge hub for cancer cell prediction.

[/expand]