Register for free to
view the full innovation profile.
We are developing a software for detection through deep learning, image recognition and 3D mapping through virtual reconstruction, for endoscopy and colonoscopy recognition.
About
We are developing a software for detection through deep learning, image recognition and 3D mapping through virtual reconstruction, for endoscopy and colonoscopy recognition. The process consists of: 1. Video export, compression, storage and playback system. This will allow all the videos to be stored locally in the Hospital permanently or for as long as it’s defined. In addition, by having a higher quality, these videos can be used later for detection by AI and the generation of the virtual scenario (Point 2 and 3). 2. Application of image recognition, artificial intelligence and deep learning for detection: development based on the artificial intelligence system with a proprietary neural network, combined with image recognition. The system will learn from the diagnosis made by the specialists and, based on this learning, it will automatically detect possible irregularities. 3. Virtual reconstruction from video format: this development consists of two sub-developments. a. Mapping: the system will interpret the shape and position of the digestive tract and the colon and will indicate, within it, where the detected irregularity is located. B. Reconstruction by means of vectors of the video content: the selected videos will be reinterpreted by the software by means of extrapolation (obtaining information from the same space from different angles or images) and obtaining information from libraries, creating a virtual vectorial reconstruction of the selected content. The main advantage is that the specialist will be able to see a specific point of the test from any perspective. In addition, when vectors are generated within this scenario, the system will recognize their proportions. The content will be accessed from a computer, through an interactive player, allowing the specialist to view the test from different angles. It could also be seen in Virtual Reality. Virtual reconstruction of a scan is estimated to take approximately 30 minutes.
Key Benefits
During the time of performing a colonoscopy or endoscopy, it is not always possible to detect and diagnose possible irregularities. On the other hand, the videos of this type of tests are saved only for a limited period of time and with the content compressed, in low resolution. Both things cause the need for specialists to repeat this type of test too often. Several drawbacks related to the current resulting content have been detected: - It is not always possible to record the entire test sequence. - Often cuts off in the middle of a scan. - The video can only be viewed again compressed, in low resolution. - If in a few weeks this content has not been exported off the server, it is removed from the system. - The current software does not allow you to export the videos to any other format or access the original material. When it comes to the diagnosis itself, this type of explorations also presents several challenges: - It is not always possible to observe the irregularities. - It is difficult to detect the size of the polyps, it has to be estimated. - It is not possible to detect exactly the situation of the irregularity. Our software will give a solution to all this drawbacks bringing great benefits to this area. Also to consider that the same software could be further developed for its application in other medical areas.
Applications
The result of the project should allow: 1. To store the videos of the tests for a longer time, occupying less space and permanently, thanks to the proprietary compression system. 2. That the stored videos have a better resolution for later viewing. 3. For professionals to access this content a posteriori to generate the detection and/or the virtual scenario. 4. That the system, through deep learning, proposes possible irregularities to professionals that they can confirm and revoke. 5. That the professional can see a reproduction of the patient's upper digestive tract and colon where the anomaly has been detected from different perspectives and to be able to locate the anomaly.