A Geometric Method for Cancer Tissue Extraction in Thermal Video

Early detection of cancer greatly increases the prospects for successful treatment. Most of the existing methods for early cancer screening involve risks, particularly exposure to radiation. Others rely on the expertise of the morphological or cytological diagnostician resulting in many false negatives.
HT BioImaging develops innovative technology to aid in detection of cancer in its initial stages with no radiation risk to the patient and without the need for an expensive analysis stage. This technology uses thermal video in the analysis stage.
The goal of this project is to identify cancerous areas by using non-linear dimensional reduction on thermal videos. During the project, we explored methods for feature extraction from a thermal video for automatic detection and classification of cancer.
As a part of this project, an algorithm for separation of cancerous areas from healthy environment was implemented in Python. The algorithm will involve technique that separates a cancerous tissue from its healthy environment in thermal (infrared) video of the tissue. The algorithm was tested on videos of several patients given by the company and exhibited good separation between the cancerous and the healthy regions.

A Geometric Method for Cancer Tissue Extraction in Thermal Video

Collaboration:
HT BioImaging