WEEK 11 FYP2: Exhibition Day

👉FYP 2 Exhibition Day
⏰9am-1pm
📌Dewan Gemilang

The students is required to set up booth at 8am. The accessors will come access students start from 9am. Alhamdulillah everything went well. Both of my accessor came early and together.

My acessor is Sir Shah Rizailli Bin Mukhtar from the Medical Electronics section and Madam Noraini, a Biomedical Engineer from Princeton Medical Centre. 

WEEK 10 FYP2: Still figure out on the Neural Networks

📌 Meet Dr. Zaki from the Electronic Section

From the data that i've collected, Dr. suggested me to add more data and parameter for training as it will be better to improve the performance. 

I also try adding K-Median filter to help enhance the image, to reduce the noise present in the blood sample image. This also will help in improving the detection process

For training, as we want to obtain the better performance and for the system to give the correct classification, train Levenberg-Marquardt is applied on the coding as it can train the networks more faster. 

https://www.mathworks.com/help/deeplearning/ref/trainlm.html;jsessionid=fff17f597d8ea3451fc8c8f950ee

https://www.neuraldesigner.com/blog/5_algorithms_to_train_a_neural_network

👉Final FYP2 Briefing by Dr. Imran Sofi
⏰ 2/10/19(Wednesday-3pm)

 

Dr. Imran remind us about the exhibition on 9/10/19

The submission for the FYP2 report is on 29/11/19

Submission of Technical paper must be through supervisor first and then to the FYP Committee.

 

👉Meet Supervisor on 4/10/19⏰9am-10am

Supervisor share few important tips on Exhibition Day. 

WEEK 9 FYP2: Understanding the architectural of Neural Networks

📔Explain the data tabulated

The data is being divided into 3 parts which is 70% for training, 15% for validation and 15% for testing. The performance of the training data can be observed by looking at the cross entropy performance in the neural networks window. 

The system achieved the better performance with it's value is near 0.00

Epoch = represent the number of iterations of the training data. 

After finish the process of detection, the classification process of training by ANN is applied to verify the validity purposed operation of detection. The benefits of training the data includes, it will be easier for the system to provide the correct output to the input. With this system, the documentation can be done specifically and the diagnosed/prognosed steps can be done faster.

WEEK 8 FYP2: Gathering data

Gather data for detection level and sorting data for ANN steps, becuase of the data provided form Hospital Ampang is limited, i tried to obtain few data on normal Red Blood Cell sample from the databank:

1. kaggle
2. UCI
3. image.hematology

The image then preprocess to obtain information from the image, make it able to be trained in the Matlab using Pattern Recognition apps.

The Sickle Cell is detected using the Form factor (Semicircular) formula. This formula detected cell which having <1.78 form factor is distinguish as sickle cell.