Description:

It's been a major challenge to identify bone age and cancer cells mainly bone cancer, but with the help of deep learning algorithms we can create a new angle to analyse bone caused problem, with these new way, doctors can easily classify and detect bone age, gender and probability of maximum height which can be achieved by any human.this method will create a great impact in medical research and development. currently there are two methods which are used my the doctor to analyze bone age and various segment with the help on X-Ray,

• Greulich and pyle (GP)
• Tanner-whitehouse (TW2)

in GP procedure involves comparing patient's radio-graph with an atlas of respective age and TW2 use scoring system based on the 20 patients with same age and gender but with help of the deep learning (using deep convolution layer) we can segment bone and analyze the important joint gap and curve of bone which result in less time span for doctor to analyze each patient bone ,because in above two procedure (GP) and (TW2) it usually take 30 min each patient time for bone analysis which is been reduced to 1-5 min per patient doctor time resulting with more accurate result.

for this experiment well be using hand X-Ray.our hand consist of three major section Phalanges,metacarpal bones and carpal bones which further increase as distal,middle and proximal this three comes under Phalanges,before feeding our data into convolution architecture we have to preprocess our data this includes making three different CNN architecture for three different bone part so that we can achieve better performance and accuracy with less amount of data set .

this architecture includes famous keras pretrained function model (VGG) key point of architecture

• VGG 64
• VGG 128
• Dropout
• ELU
• output

Conclusion

In this talk we investigated the application of deep learning convolution neural network to the problem of automatic bone age and bone cancer cell detection.

Rest will be explained in my talk

Time step will include

• 2 min on intro video on neural networks
• 3 min on architecture on neural networks
• 2 min on connection of deep learning and neural networks
• 7 min on detail analysis of bone structure and analysis
• 10 min on using deep learning and neural networks to analyse the bone structure.
• 1 min on conclusion
• 5 min on Q&A

Prerequisites:

I would recommend following prerequisites:

• basic knowledge on neural networks
• image processing using convolution neural network
• python

if audience would consist of practitioner doctor this would add a perk to talk

Content URLs:

keras VGG

propasal content

Speaker Info:

Name :- harsh Mishra,pursuing computer science engineering third year from SRM institute of science and technology , had completed more than 14 major projects on field of machine learning and artificial intelligence, and from my view , machine learning and artificial intelligence will make a major impact on human research and development.also worked under ministry of heavy transport government of India to implement great research results. IIT Madras certified machine learning certificate

Speaker Links:

-profile on LinkedIn

• Profile on GitHub