We provide a SaaS solution that can identify cancers from pathological images by utilizing artificial intelligence and machine learning. It has shown to be up to 93% accurate in early detection and classification of cancer tumors.
How We Do It
Data is central to our business and plays a vital role in feeding our AI engine to improve the quality of outcomes.
Our data science and engineering teams work with major research, commercial and government data platforms to collect enormous amounts of data that feeds our Machine Learning frameworks and AI algorithms. Our thorough and robust models and testing software ensure the quality of our system is near perfect and we are able to select and deploy the best features for each use case. Our process is fully compliant with Global Data Privacy laws.
Our data providers may include:
Our technology platform consists of the most scalable, robust and advanced global infrastructure and network combined with highly skilled data science and software development capabilities.
We are an Amazon Web Services (AWS) partner
Our Data Science and Software Engineering teams are strong
- An RGB image is input to a set of convolutional layers
- Very small receptor field filters are used for liner and non-liner transformation of input channels
- Tailored Fully-Connected layers are utilized and the last layer contains neurons that classify tumors into classes such as meningioma, glioma and pituitary
- Model is refined using Gabor filters and color blobs
- Multiple tumor image data sets containing thousands of T1-weighted contrast-enhanced images from patients where multiple tumor type slices are processed to train and test the model
- Primary Features analyzed include size, position, shape and texture of the tumor
- Datasets are continually enhanced and improved