01 / BIG DATA
Our final products are based on a Convolutional Neural Network Classifier (CNN), which has been trained by analyzing more than 10,000 different images and scenarios using a pre-process of deep learning for AUC (area under the curve) and mask removal by color frequency in a method called Laplace Pyramid for the pre-processing training stage
02 / DEEP LEARNING
The built CNN models were trained to detect several tags such as: a general polyp, a flat polyp and area of the cecum.
For medical data classification, sensitivity (true positive rate) and specificity (true negative rate) are more reliable than accuracy (rate of successful detection).
03 / PRODUCT
Computer aided polyp detection systems can reduce polyp mis-detection rate and assist physicians in finding the most important regions that need to be analyzed. This can be achieved as an offline check or during a real-time procedure. Such a system can support the diagnosis procedure by detecting polyps, classifying polyps and generating a detailed report with regards to any part that requires further and more detailed examination.
The Sleuth Able to Detect Gastrointestinal Polyps and Assist in Early GI Cancer Discovery
Just like in the days of Sir Arthur Conan Doyle's famous story of the legendary detective Sherlock Holmes and his crime solving abilities, a story of a detection system has entered the Gastrointestinal medical field. A system able to sense cancerous polyps and support the prevention of Gastrointestinal Cancer, allowing physicians to achieve a better diagnosis and by extension saving the medical industry millions of dollars in healthcare costs.
It is a known fact that Gastrointestinal Cancer is one of the leading, if not the most common of all, cancers. It is also a known fact that if diagnosed early, this form of cancer can be cured. The slow growth of cancerous cells in the intestinal tract means that when patients report problems, it is often when cancerous cells have developed, and cancer has spread. So early detection is key in prevention of this disease.
Men and woman ages fifty and over (depending on family history, this may be earlier) are required to undergo a colonoscopy. If the procedure is negative, a person is required to repeat the procedure every ten years.
But let’s look at what happens in the examination. Physicians are human. And as humans, they too, may make mistakes or simply overlook certain sections of the intestinal tract during an examination.
Enter our detective. Sherlock. Unlike the fictional character, this Sherlock is very real. Designed to help detect cancer and to save lives.
Sherlock is a software engine able to detect polyps in the gastrointestinal tract using a unique image analysis and deep learning algorithm that through a process of identifying and studying thousands of polyps can provide physicians with an additional means of diagnosis during an examination.
Sherlock is not the only means of detection available to physicians. However, it is designed to provide physicians with a deep learning system that has incredible calculation abilities that improve the diagnostic capacities of physicians whilst maintaining excellent patient care.