Offering high sensitivity, specificity, and anatomical penetration, MRI is the best diagnostic technology for detecting prostate cancer. However, despite the high quality and fast delivery of MRI imaging, the challenges of MRI interpretation lead to significant issues. Even the best trained radiologists with hundreds of cases studied miss cancerous lesions (false negative), under-estimate primary lesions (leading to inaccurate prognoses), and/or over-estimate or misinterpret (false positive) pathologies in the statistical range of 15-35% of the time. Therefore, MRI screening for prostate cancer has not been adopted as the standard of care.
We believe ProstatID will overcome this clinical hurdle by eliminating the visualization limitations of radiologists using complex, wholistic AI algorithms.
ProstatID's conceptual AI algorithm was first developed by code writers physicians at the National Institutes of Health (NIH). The algorithm overcomes the visual and cognitive limitations of humans by learning how to best interpret the MRI data. In other words, the program can assess visible features such as brightness and gray scale but also those invisible to the human eye such as pixel by pixel texture, 3D voxel texture, and infinite feature weighting testing and validation.
The first revision AI software platform of ProstatID was tested on 270 men with prostate cancer symptoms, leading to the findings published in the peer-reviews Journal of Medical Imaging (4(2), 024506 (Apr-jun 2017)). You can read their findings here:
Detection of prostate cancer in multiparametric MRI using random forest with instance weighting
Automatic magnetic resonance prostate segmentation by deep learning with holistically nested networks:
This prototype technology was then exclusively licensed to ScanMed, LLC, a world-leading MRI technology company with over a dozen patents, including the world's first Prostate MRI antenna, the PROCURE, to develop and bring to market.
To date, ScanMed, dba Bot Image, Inc, has acquired over 1700 proprietary data sets, most with radiologists interpretations, segmented prostate gland image sets, pathology laboratory results, correlation data, and first-pass AI interpretations. This is the largest collection of data sets in the industry and will work to further improve the accuracy of ProstatID's AI image interpretation capabilities. ProstatID also seamlessly integrates into the radiologist workflow and we predict as more studies are interpreted the better the technology will become with the potential of our learning algorithm.