DevMesh

Brain Tumor Classification and oneAPI

Anushka Panda

Anushka Panda

Kanpur, Uttar Pradesh

0 0
  • 0 Collaborators

Existing literature uses CNN techniques and multiple variations and derivatives of the same to detect and classify tumors. This project explores the opportunity to: Utilise oneAPI to analyse the available datasets Utilise oneAPI libraries to try and implement the current existing methods for tumor c ...learn more

Project status: Under Development

Artificial Intelligence

Intel Technologies
oneAPI

Code Samples [1]

Overview / Usage

What’s a tumor?

Tumor is an abnormal tissue mass formed when cells divide uncontrollably or show an abnormal life cycle. It may be benign (localized) or malign (spread to other regions). Malignant tumors are cancerous in nature and may be fatal if medical intervention is not granted as soon as possible.

Why does it take so long to diagnose?

From patient’s awareness to clinical diagnosis, detection of a tumor may take a long time based on the time lag in response at each level. Patient awareness issue may be solved by educating the population, but clinical diagnosis may often be delayed if the size of the tumor is microscopic at a given point in time. For instance, a brain tumor in an MRI may not be detected quickly if its size is too small to be visible to a radiologist.

Not to mention, radiological diagnosis is also very expensive.

Thus, tumor detection with the help of AI will not only help in faster diagnosis, but also be a step towards Goals 3 and 9 of the UN’s 17 Sustainability goals

Methodology / Approach

Firstly, the project tries to apply the given oneAPI's sklearnex library to patch sklearn for optimisation, in order to build a kNN classifier for the tumors.

Secondly the project tries to use the tensorflow framework for oneAPI to try and optimise the model's training time.

Repository

https://github.com/apanda1001/Machine-Learning-using-oneAPI/blob/main/intel_oneapi.ipynb

Comments (0)