The project 'Tumor detection using thermal imaging' makes use of it's high quality CNN architecture to determine the presence of tumor in living tissues by making use of it's thermal imaging. It is based on the biological concept that tumor cells are warmer than surrounding tissues and it is adiabatic in nature. So the tumor will appear as a brighter spot in thermal image of that tissue. We will use a 6 layer CNN to classify the presence of tumor and to differentiate it from the temperature change occurred due to other causes. Our system will mark the spots containing tumor along with a confidence score to determine whether the spot is actually a tumor or not. Our project will reduce the need of using expensive CT scans and also give the patient an early warning so that it can be treated at early stages.
1.Programming language: Python 3.6.4
2.IDE: Anaconda Spyder, Jupyterlab
3.Frameworks:
i.Tensorflow for dataflow programming
ii.Keras for defining layers
iii.OpenCV for object detection tasks
4.Additional Softwares:
i.NVIDIA CUDA 9.0 for GPU acceleration
ii.Optima for code optimization.
Dataset can be found here: https://drive.google.com/open?id=1vhm827aOBXF9apU8ByKYJQCx7N0Se5rf