MammoVisionTM is an artificial intelligence based computer vision platform for analysis of mammograms. The system provides an initial analysis of mammograms to assist radiology professionals in making their diagnosis faster and reduce human errors. To make it work we had to dig deep into our knowledge of computer vision and AI, come up with tools to manually annotate thousands of public datasets to teach our CNN in a short span of time.


Artificial Intelligence is hyped as the next big thing that would change the way we do things fundamentally. A statement yet to be proven. However due to the progress achieved in the recent years we see evidence to strongly believe that AI is at an inflection point. We could see AI based smart and advanced tools that would work alongside radiology professionals to help them achieve peak performance and make diagnosis easier, efficient and accurate than ever before.


With over 50M mammogram screenings a year in the US alone there is a huge volume to handle. The other aspect is the need for high accuracy. To tackle these problems one method that is emerging is to use Artificially Intelligent smart tools to assist professionals to enable them handle higher volumes while enhancing accuracy.


We have developed MammoVisionTM which is an artificial intelligence based computer vision platform for analysis of mammograms and X-rays. We have created algorithms to detect and classify features of a mammogram image. Our system self learns and keeps getting better over time. Currently after training with 1000’s of mammograms we have obtained detection capabilities that closely matches trained human classification standards. We have a beta version that provides initial analysis of mammograms and can assist professional radiologists in making their diagnosis faster.

The platform is available on the internet. Any mammogram image can be uploaded to our server via our secure portal. Once images are received our patent pending algorithm analyses and responds with a marked-up image with analysis results.


After a mammogram review the system classifies features it finds into malignant, benign, calcifications lesions with a confidence score. A higher confidence score indicates higher probability that system is accurate in its result in comparison with previous images it has seen and learned.

As depicted in the third image from gallery MammoVision system provides a training accuracy of 94%. This was achieved with training dataset of about only 8000 manually marked up mammogram images. By this we mean that out of 100 new cases reviewed the system detects 94 cases correctly as shown on the right side. Once the image is analyzed the system can mark circles around suspected areas.


We are looking for strategic partnerships to develop the product further. In return we provide early access to our product and free commercial licenses no cost to all our partners. We are happy to engage and showcase our demo if you are further interested.

Let’s Create Big Stories Together!

Transform your vision into reality. Contact us for expert software development services.