You are here

Automated segmentation of tissue in MRI to aid diagnosis of breast cancer

Researchers from the Indian Institute of Technology, Delhi and the Fortis Memorial Research Institute, Gurgaon, have devised a new way to automatically differentiate between inner and outer breast tissue using Magnetic Resonance Imaging or MRI in breast cancer patients.

Breast cancer occupies the top spot for cancer affecting women world over. Due to improvement in diagnostic and treatment techniques, the survival rate for this kind of cancer is higher compared to many others. MRI is the most common technique used to diagnose breast cancers. It allows a physician to get detailed images of a patient’s anatomy using strong magnetic fields. In an MRI image along with the breast tissue other organs of the body like pectoral muscles, heart and lung are also seen. To add to the difficulty of a physician attempting to diagnose the presence of a cancerous tumour, the breast tissue itself is composed of many kinds of tissue. Hence an expert needs to study this image, to understand where exactly a cancerous tumour is present.

This process where a doctor must decide, leads to subjectivity. In their study the researchers have tried to automate the process to reduce the subjectivity involved in the diagnosis. This decision-making process is crucial as the results of the MRI are used to make post treatment evaluations as well as for planning radiotherapy treatments.

For their study the researchers used MRI data from 30 patients. This data was then subjected to manual segmentation of the inner and outer breast tissue as well as the automatic segmentation devised by the researchers. The automatic segmentation of the MRI was carried out by algorithms created by the researchers using MATLAB. The results of the automation were then validated by two experienced radiologists.

The results of the study show that the automation technique was able to distinguish between different kinds of tissue in outer and inner segments of the breast with accuracy. Apart from 2-3% of mismatch, the segmentation was seen to be accurate when validated by experienced radiologists. Other factors such a breast density and tumour density were also automatically calculated using the technique.

“The proposed approach enables automatic estimation of breast density. Breast density estimation plays a significant role in clinical aspects. This estimation could be used in follow-up studies (after chemotherapy) to detect small changes in breast density. This estimation can be taken into account for various studies such as epidemiological and parenchymal (morphological distribution pattern of the fibro-glandular tissue) etc.”, sign off the researchers.