Scribd
Upload
83 views3 pages

Presentation Classification and Segmentation of Tuberculosis

According to the WHO, tuberculosis kills nearly 1.8 million people worldwide each year. Diagnosis of tuberculosis in third world countries currently relies on conventional methods like blood tests and sputum analysis that can take 1-2 weeks. The goal of this project is to develop an AI-assisted tool that can classify x-ray images and detect tuberculosis lesions faster using devices like the Raspberry Pi and Nvidia Jetson Nano. The team will acquire x-ray images, use deep learning models for detection, classification, and segmentation to diagnose tuberculosis in less time than conventional methods. They will work in two phases - detection and then classification or segmentation.

Uploaded by

saissi1701
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
83 views3 pages

Presentation Classification and Segmentation of Tuberculosis

According to the WHO, tuberculosis kills nearly 1.8 million people worldwide each year. Diagnosis of tuberculosis in third world countries currently relies on conventional methods like blood tests and sputum analysis that can take 1-2 weeks. The goal of this project is to develop an AI-assisted tool that can classify x-ray images and detect tuberculosis lesions faster using devices like the Raspberry Pi and Nvidia Jetson Nano. The team will acquire x-ray images, use deep learning models for detection, classification, and segmentation to diagnose tuberculosis in less time than conventional methods. They will work in two phases - detection and then classification or segmentation.

Uploaded by

saissi1701
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

PRESENTATION OF OUR

WORK
According to the World Health Organization (WHO), tuberculosis kills nearly 1.8 million people
worldwide each year.

Tuberculosis (TB) is an infectious disease caused by a mycobacterium that spreads through the air to both
children and adults. It most often affects the lungs but sometimes also affects other organs and can
progress to death if left untreated or detected in a late stage.

On average, more than 10 million people contract it each year. More than 95% of deaths from
tuberculosis occur in low- and middle-income countries, including Algeria which is fighting it since the
80’s.

Nowadays, In third world countries including Algeria, the diagnosis of tuberculosis is done through
conventional methods. Diagnostic results are obtained from conventional methods such as blood, culture,
sputum and biopsies. They are tedious as well as take long time like 1-2 weeks or maybe even more.
Therefore, to lower the detection time and raise the accuracy of diagnosis we can take advantage from the
the rise of artificial intelligence that was democratized lately due to the evolution of hardware acceleration
such as GPU’s especially Nvidia solutions.

Our goal is to come with a solution for doctors especially radiologists to help them and assist them to
come with a diagnostic in less time than conventional methods, to treat more people quickly with reduced
costs.

Using the development kit, we will take advantage of each component available to realize a complete tool
to classify whether an x-ray image represents a tuberculosis case or not, and to detect the affected zones if
the x-ray was classified as a tuberculosis case in the first step.

First, we will use the Raspberry pi camera module to acquire an image representing the x-ray, this image
will be processed by our first deep learning model to extract the x-ray, then we will use the latter as an
input to our second neural network that will classify whether the x-ray represents a tuberculosis case or
not, finally if the x-ray represents a tuberculosis case we will segment the affected zone. We will use the
NVIDIA Jetson Nano and the Intel Neural Compute Stick 2 (NCS 2) to run the deep learning models
that we will need for the accomplishment of our task .
The following illustration shows the steps of our work :

Illustration of our workflow.

To realize this project, we will go through 2 phases:

Phase 1: All members of the team will work together to realize the first task (Detection of the x-ray on the
image captured by the Raspberry pi camera module)

Phase 2: We will form two teams, each one of them will work on a specific task (classification or
segmentation)

-The first team will be : -BERKANE Ania.

-AISSI Mohamed Salim.

-The second team will be : -HALILALI Melissa Rania.

-HAMMOUCHE Yanis.
Gantt chart:

Task December January February Mars April May


Planning X
Research X X
X-ray X X
detection
Classification X X
Segmentation X X

You might also like