AI-powered Blood Tests

October 2019

Body check-ups routinely include blood tests, a key component of which is an inspection of the white blood cells. Traditional blood tests involve staining and labelling, which are time-consuming and labour-intensive. While newer automated machines speed up the process, the difficulty of moving them around and the high costs remain an issue.

These problems are going to be solved by CUHK’s AI-enabled portable quantitative phase microscope. Being the first of its kind, this microscope will make blood tests cheaper, faster and readily available at general clinics and even in underdeveloped regions.

The microscope is the work of a team led by Prof. Zhou Renjie of the Department of Biomedical Engineering. Its main advantage is its artificial intelligence which allows delegating the task of identifying white blood cell types. What makes this possible is a Quantitative Phase Microscopy (QPM) perfected by Professor Zhou’s team, by which one can now obtain precise results using cheap and compact equipment.

‘In the past, researchers have tried to combine artificial intelligence with traditional blood testing methods but in vain, because it is difficult to distinguish cell images. Through our high-precision QPM technology, we can effectively combine it with the deep learning technology of artificial intelligence,’ Professor Zhou said. He explained their AI model can now learn cell features from high-quality images yielded by the enhanced QPM and in turn, differentiate between the many types of white blood cells in a healthy volunteer’s blood sample.

The microscope weighs less than five kilograms and is around the size of a briefcase, meaning it can easily be carried to any place for use. Also, the fact that it takes labelling out of the procedure means reagents are saved. Much human effort is saved since the cells are now counted and classified by a computer, which offers results in a matter of minutes with over 90% accuracy.