Lunit
Back to List

What do the medical journals say about AI-powered mammography?

Feb 18, 2021 — 4 min read

Bokyeong Woo

“AI? I am hearing about AI everywhere I go.”


“But has it reached the level where I can use it in my daily practice? Can I really rely on AI analysis results for breast cancer detection?”


“Lunit AI for mammography? Has the software performance been validated?”


“What does the clinical evidence say?”




We are going to answer your questions by reviewing the key findings of each study published in major peer-reviewed journals.


Click to watch the animated video about Lunit INSIGHT MMG, an AI solution for mammography







Image for post




[Question 1]

Can Lunit AI solution really improve my mammography interpretation?

Yes.


Image for post


According to the study published in The Lancet Digital Healthusing Lunit AI solution, radiologists can improve their diagnostic performance for breast cancer. In particular, general radiologists significantly improved their diagnostic performance, at a level up to that of breast specialists.







Image for post

Lunit AI algorithm also demonstrated high sensitivity in detecting early breast cancer such as T1 and node-negative breast cancer.







Image for post

This study demonstrates that Lunit AI solution can be used as an effective diagnostic support tool in various clinical settings.







Image for post




[Question 2]

Can I rely on AI results when triaging cases?

Yes.


Image for post

According to the study published in The Lancet Digital HealthLunit INSIGHT MMG can safely rule out normal cases without missing any breast cancers, and detect more cancers which are normally identifiable in the next round of screening.







Image for post

An abnormality score is the AI’s confidence level on the existence of detected breast cancer.







Image for post

When deployed as a single reader without radiologists, Lunit INSIGHT MMG successfully triaged cases with low 60% of abnormality scores without missing any screen-detected cancer.







Image for post

On the contrary, does Lunit AI solution accurately detect more breast cancers on mammograms with high abnormality scores? 32~35% of the entire cancer cases that were diagnosed negative by radiologists, but had the top 2~5% of AI scores, were detected to have cancer in the next round of screening.







Image for post

This finding suggests you two things:


First, you can rely on Lunit AI solution in triaging normal cases and focus on abnormal cases, ultimately reducing your workload by more than half in mammogram interpretation.


Secondly, you can use the AI result to reduce false-negative recalls, and order the next round of exams such as ultrasound and biopsy for the patients who are highly suspicious of having breast cancer.







Image for post

[Question 3]

Why do I have to choose Lunit INSIGHT MMG over other AI solutions?

Well, Lunit AI solution is simply the best.







Image for post

According to the study published in JAMA OncologyLunit INSIGHT MMG outperformed other commercially available AI solutions in breast cancer detection.







Image for post

According to the study, the combination of first-reader radiologists and Lunit INSIGHT MMG outperformed not only the first-reader and second-reader radiologists but also the double reading by radiologists.







Image for post

This outcome shows the potential value of Lunit INSIGHT MMG as a second reader, which ultimately enables radiologists to allocate their time and resources more efficiently.







Image for post

Online demo website (http://insight.lunit.io)






Try Lunit AI solution now at insight.lunit.ioOnce you upload DICOM images, you will get the AI results within seconds.

AIArtificial IntelligenceBreast RadiologyLunitMammographyRadiology

More from Blog