TACESM: Textural Analysis of Contrast Enhanced Spectral Mammography
Research type
Research Study
Full title
Textural Analysis of breast lesions on Contrast Enhanced Spectral Mammography: a feasibility study
IRAS ID
266560
Contact name
Sarah Savaridas
Contact email
Sponsor organisation
University of Dundee / NHS Tayside
Duration of Study in the UK
1 years, 8 months, 29 days
Research summary
Summary of Research
Breast cancer remains the most common malignancy in women in Scotland, accounting for 29% of all female cancers in 2015, and 13.5% of female cancer deaths in 2016. It is accepted that early detection and diagnosis can improve patient outcomes. Within the UK, breast lesions considered to be of at least indeterminate suspicion on either imaging or examination are biopsied. Management and treatment is based on these core biopsy results. There are two downsides to this practice. Firstly, it results in many benign biopsies which may be avoidable if malignancy could be excluded on imaging. Secondly, due to the heterogenous nature of breast tumours, core biopsy may not give a true representation of the tumour grade and immunophenotype due to the limited tissue sampled. As prognosis and response to treatment is dependent on this information, over or under treatment of disease may occur. In the era of personalised medicine, it is important to identify the specific breast cancer subtype and tailor treatment accordingly. If this were possible, using analysis of imaging, it would be beneficial for the patient and treating team.
Textural analysis (TA) is a technique for evaluating the appearance, position and pattern of pixels according to their grey-level intensity within 2D digital images. Research has shown that when this is applied to breast MRI images it may be possible to distinguish benign from malignant lesions and differentiate histological tumour types. Contrast-enhanced spectral mammography (CESM) is a new functional imaging technique, potentially offering a better tolerated and cost-effective alternative to breast MRI. We seek to establish whether it is feasible to apply TA to CESM images and, if so, whether it is possible differentiate benign from malignant lesion and between malignant lesion histotypes.Summary of Results
This study involved analysis of contrast enhanced mammogram (CEM) images. CEM is a new x-ray technique that uses contrast dye to highlight areas of the breast with an increased blood supply, such as cancers. Textural analysis is a technique whereby computer software finds patterns within images, in the case of this study, with in CEM images. Textural analysis may improve how accurate CEM is at finding cancers. This study was performed to see if textural analysis worked on CEM images and if cancer and non-cancer masses could be accurately told apart.CEM images were teken using machines from two different companies; Hologic and GE equipment. The computer software used was called 'MaZda analysis software'. A doctor drew around the different masses on the CEM image, so the computer know were to look for patterns. There were two different ways to draw round - freehand and using an oval (ellipsoid) shape. The patterns were then used to build a computer model to tell which masses were cancer, and which were not.
The computer models were very accurate and able to tell whether a mass was cancer or not in more than 9 out of 10 cases. These results will help further developments aimed at producing a widely accessible radiomics model for clinical use.
REC name
West of Scotland REC 1
REC reference
19/WS/0113
Date of REC Opinion
9 Aug 2019
REC opinion
Favourable Opinion