University of Pittsburgh researchers have developed an automated algorithm which identifies epithelial nuclei for cancer diagnosis.
Description
An important advancement needed in the treatment and diagnosis of cancer is the unbiased and rapid identification of disease biomarkers. Using tissue samples of Barrett’s Esophagus from healthy individuals and patients with cancer, scientists at the University of Pittsburgh have developed a computational model which unbiasedly identified the nuclei in the data set and correctly labeled them as epithelial or non-epithelial. Automated cell type-specific detection gained through this algorithm has not only aided in the diagnosis of esophageal cancer, but also in the development of novel biomarkers and strengthened existing algorithms used for disease detection.
Applications
1. Diagnostics
2. Novel biomarker identification
3. Cancer detection and treatment
Advantages
The current benchmark in diagnosing various cancerous tissues relies upon the expertise of highly skilled pathologists who employ traditional techniques to meticulously scrutinize cellular nuclei based on a predefined set of criteria within biopsy samples. This process is both laborious and time-consuming, demanding pathologists to painstakingly discern and categorize cells within the samples. This new computational modeling developed by University of Pittsburgh allows for unbiased identification of diseased cells. The automated nature of the algorithm allows for increased speed (180 seconds per image compared to several minutes required by a highly trained researcher to manually outline each nucleus in an image) and for samples to be run at any time or date. Therefore, the algorithm leads to higher-quality nuclei detection for diagnostic purposes.
Invention Readiness
A software has been developed and tested on human Barrett’s Esophagus tissue from both healthy individuals and patients with cancer. This technology is patented and ready to be used in larger cohorts. Software which requires further optimization for cancers and other diseases.
IP Status
https://patents.google.com/patent/US9626583B2
