Many women especially in the rural parts of the world, don't have access to healthcare and knowledge about various diseases. There must be widespread knowledge about all types of cancers that affect women.

All lives are precious but a woman's life is more precious than any other. No woman should die of cervical cancer, as it is TREATABLE if detected in early stages.

We have developed a tool that helps to detect cervical cancer early. This tool will help doctors for fast diagnosis and detection of cervical cancers and also reduce human errors because of the vast amount of data it is trained on.

Our tool ka[AI]li facilitates the advancement of the process of early identification of cervical cancer using multi celled PAP smear images.

It develops in a woman's cervix (the entrance to the uterus from the vagina).
Although most infections with HPV resolve spontaneously and cause no symptoms, persistent infection can cause cervical cancer in women.

It is the fourth most common cancer in women.

Effective primary (HPV vaccination) and secondary prevention approaches (screening for, and treating precancerous lesions) will prevent most cervical cancer cases.

How we built it

We first extracted the abnormalities out of the images using OpenCV and performed various morphological operations on the images and then based on the abnormalities, we classified them as cancerous or noncancerous.

We used the Support Vector Machines (SVM) to classify the images based on their shape and count of the cancerous clusters or cells.

We used OpenCV, Pillow and tensorflow to build the classifier model.

The web app has its backend in flask and front end in html, css and js.

The web app is deployed with Microsoft Azure App Services.

Accomplishments that we're proud of

We are happy we have been able to build something which has the potential to make early detection of to all thus saving lives of women across the world.

What we learned

We explored several image processing techniques using OpenCV and SciKit Image to classify the PAP smear images. We used Azure for the first time to deploy a Flask web app to the cloud.

What's next for ka[AI]li

We plan to host this in production after publishing our research on this.

Instructions to run from terminal

First clone the repo :

git clone https://github.com/Soumi7/HackV21.git
cd HackV21
python3 app.py

Instructions to run after installing VS Code

code HackV21

Main actions are commit, push and pull.

  • Go to the three connected dots on the left pane

  • For making changes, these are the steps :

    • Make changes
    • Save the file
    • Go to the three connected dots on the left pane, i.e github
    • The files with changes will appear
    • Move your cursor on the file, see a + symbol, click on it to stage the changes
    • Now write a commit message and click on the tick on top to commit
    • Now these commited changes have to pushed so that they are updated on github online
    • Go to the three connected dots on the top right corner of the pane.
    • You will see push, click on it
    • It will ask your name and password, enter them.
  • Now for pulling changes that someone else made to the repo :

    • Go to the three connected dots on the top right corner of the pane.
    • Click pull

Built With

Share this project: