-
-
Example #1 of Gene mutation detected on Genetic Test Report image
-
Example #2 of Gene mutation detected on Genetic Test Report image
-
Auto-generated PDF report of example Gene mutation
-
Terminal image showing PASS results for Gene detection in related image
Inspiration
The inspiration for this project came from UofTHacks Restoration theme and Varient's project challenge. The initial idea was to detect a given gene mutation for a given genetic testing report. This is an extremely valuable asset for the medical community, given the current global situation with the COVID-19 pandemic. As we can already see, misinformation and distrust in the medical community continue to grow in popularity, thus we must try to leverage technology to solve this ever-expanding problem. One way Geneticheck can restore public trust in the medical community is by providing a way to bridge the gap between confusing medical reports and the average person's medical understanding.
What it does
Geneticheck is a smart software that allows a patient or parents of patients with rare diseases to gather more information about their specific conditions and genetic mutations. The reports are scanned through to find the gene mutation and shows where the gene mutation is located on the original report.
Genticheck also provides the patient with more information regarding their gene mutation. Specifically, to provide them with the associated Diseases and Phenotypes (or related symptoms) they may now have. The software, given a gene mutation, searches through the Human Phenotype Ontology database and auto-generates a pdf report that lists off all the necessary information a patient will need following a genetic test. The descriptions for each phenotype are given in a laymen-like language, which allows the patient to understand the symptoms associated with the gene mutation, resulting in the patients and loved ones being more observant over their status.
How we built it
Geneticheck was built using Python and Google Cloud's Vision API. Other libraries were also explored, such as PyTesseract, however, yielded lower gene detection results
Challenges we ran into
One major challenge was initially designing the project in Python. Development in python was initially chosen for its rapid R&D capabilities and the potential need to do image processing in OpenCV. As the project was developed and Google Cloud Vision API was deemed acceptable for use, moving to a web-based python framework was deemed too time-consuming. In the interest of time, the python-based command line tool had to be selected as the current basis of interaction
Accomplishments that we're proud of
One proud accomplishment of this project is the success rate of the overall algorithm, being able to successfully detect all 47 gene mutations with their related image. The other great accomplishment was the quick development of PDF generation software to expand the capabilities and scope of the project, to provide the end-user/patient with more information about their condition, ultimately restoring their faith in the medical field through a better understanding/knowledge.
What we learned
Topics learned include OCR for python, optimizing images for text OCR for PyTesseract, PDF generation in python, setting up Flask servers, and alot about Genetic data!
What's next for Geneticheck
The next steps include poring over the working algorithms to a web-based framework, such as React. Running the algorithms on Javascript would allow the user web-based interaction, which is the best interactive format for the everyday person. Other steps is to gather more genetic tests results and to provide treatments options in the reports as well.
Log in or sign up for Devpost to join the conversation.