An Enhanced Artificial Intelligence Breast MRI Interpretation System

Breast Cancer Clinical Trial

— IntelliScan  

Official title:

A Comparative Single-centre Study to Evaluate an Enhanced Artificial Intelligence Breast MRI Interpretation System in Women Over 20 With Breast Lesions

Clinical Trial Details — Status: Not yet recruiting

Administrative data

• NCT number: NCT03829423
• Other study ID #: 4901
• Status: Not yet recruiting
• Phase: N/A
• Start date: April 2019
• Est. completion date: July 2020

Study information

• Verified date: February 2019
• Source: Teesside University
• Contact: Jamil Kanfoud, M.Eng.
• Phone: +44(0)01223940
• Email: jamil.kanfoud[@]brunel.ac.uk
• Is FDA regulated: No
• Study type: Interventional

Clinical Trial Summary

Interpretation of breast MR images is a very time-consuming process and places a great burden on breast radiologists. This project aims to develop a technical solution that addresses this healthcare challenge by developing a system that is able to automatically interpret breast MR images in order to aid the radiologist in their diagnosis.

Description:

Breast cancer is the most common type of cancer in women worldwide, with nearly 1.7 million new cases diagnosed in 2015. In the UK, one in five cases of breast cancer results in a fatality. The IntelliScan project aims to develop a technological solution that addresses a significant healthcare challenge. IntelliScan will develop a software system that will be able to interpret breast MR images automatically in order to identify potential breast cancers.

Regular MRI screening of the breast is offered to women from the age of 20, who are at higher risk of developing breast cancer. MR image sequences provide a large amount of information to the radiologist and the interpretation of images is a manual process, which is very time consuming. The high number of women eligible for MRI screening combined with the amount of data provided by MRI scans places a great burden on healthcare systems. Therefore, automatisation of this process would greatly relieve this burden and also has the potential to provide more accurate diagnoses.

In this first study, the system's user interface as well as the algorithm will be developed using existing MRI scans. Existing MRI scans with known breast anomalies will be used to develop the decision-making basis for the algorithm. The system will then be tested using existing MRI scans without information about possible anomalies and results will be compared to results from the software system currently in use. In addition, the user-friendliness of the system's user interface will also be evaluated.

Recruitment information / eligibility

• Status: Not yet recruiting
• Enrollment: 1526
• Est. completion date: July 2020
• Est. primary completion date: January 2020
• Accepts healthy volunteers: No
• Gender: Female
• Age group: 20 Years and older

Eligibility

Inclusion Criteria:

• Breast MRI scans

• MRI examinations undertaken at partner NHS Trust in the UK

• MRI examinations undertaken on the MRI system currently installed at partner NHS Trust site (since 2008)

Exclusion Criteria:

• Incomplete breast MRI datasets

• Breast MRI without lesions

• Breast lesion on MRI not biopsied

Related Conditions & MeSH terms

• Breast Cancer

Intervention

Diagnostic Test:
• Breast MRI interpretation
Analysis and interpretation of breast MRI sequences by a specially developed breast MRI interpretation algorithm

Locations

Country	Name	City	State
n/a

Sponsors (3)

Lead Sponsor	Collaborator
Jamil Kanfoud	Brunel University London, First Option Software Ltd.

Outcome

Type	Measure	Description	Time frame	Safety issue
Primary	Sensitivity/specificity of breast interpretation algorithm	Sensitivity and specificity of the information provided by the breast interpretation algorithm to be above 90% and 70%, respectively	1 year
Secondary	Time required for diagnosis	The time required to arrive at a diagnosis using IntelliScan should be less than using manual procedures	1 year
Secondary	User-friendliness of IntelliScan system	Obviousness score for categorisation of beast lesions (0 [not obvious] to 100 [extremely obvious]); ease-of-use score for IntelliScan system (0 [not easy to use] to 10 [extremely easy to use])	1 year