System for the recognition of tumour masses on mammographic images

DeepLook is a Computer Aided Detection (CAD), based on Deep Learning algorithms, for the automatic detection and classification of tumour lesions in 3D mammography images (digital breast tomosynthesis). he purpose of DeepLook is to assist the senologist in locating suspicious lesions.

How does it work?

Digital tomosynthesis, or 3D mammography, is able to separately visualize objects placed at different depths by capturing 10-20 low-dose two-dimensional projections for different angles of the x-ray tube around the breast. The acquired data are reconstructed in a series of thin layers at high resolution.

The DeepLook system is trained on one set of images in which the doctor has already identified the presence of tumour lesions; then a test is carried out on another set of images to verify that the system is able to recognize the tumours that the doctor has diagnosed.

Once the system is trained and tested, it can become operational and help the senologist identify suspicious tumour masses on mammographic images.

Applications

• Early diagnosis of breast cancers.
• Screening for cancer prevention.

Advantages

DeepLook can help the senologist in the detection of early-stage cancers, contributing to the prevention and treatment of breast cancers