GSOC 2020 project idea 29: DIPY: Deep learning based MRI registration

Thanks alot @ShreyasFadnavis. I would definitely look into this issue

Hi @ShreyasFadnavis and @bramsh, I am Atharva Sundge a self taught programmer from India.I am familiar with Python from last 2 years and genuinely interested in Deep Learning specially CNN,so done (played) a lot with TensorFlow and also having strong background in Maths and also undrestanding.
I am interested in the project named ‘DIPY: Deep learning based MRI registration’ .I will go through all links/steps shared above by @ShreyasFadnavis and @bramsh.
Looking forwards for suggestion.
Thank you.

Hi @ShreyasFadnavis and @bramsh, I am Aryansh Omray; currently, a sophomore pursuing Computer Science & Engineering in Indian Institute of Technology, Varanasi, India. I am interested in working on this project in the GSOC program. I have a strong mathematical background and worked with Tensorflow in previous projects.
I have gone through the papers and the tutorials that were shared on this forum till now and tried to gain an understanding of the topics, including Diffeomorphic Registration and some image reconstruction models in DIPY module. I have also studied some 2D approaches in computing deformation field.
Could you please post some resources on the 3D computation of deformation field.
Thank you.

1 Like

Hi @aryanshomray ! This is where I learnt most of the stuff from:

A review of deformable surfaces: topology, geometry and deformation

It gives a very good starting point to understanding things! Does this help?

Hi, my name is Adit Kotwal and I am currently a final year computer engineering student from Mumbai. I found this project idea quite interesting and would love to make some contribution. I see there is a lot of documentation in this thread and will try to implement a deep neural net.

Thank you @ShreyasFadnavis for sharing the resources. I have gone through the paper you have provided. I have started going through the code base of DIPY module. I have started designing a NN based architecture for the project. I just wanted to ask for links to some medical datasets that are relevant to the task, so that I can train and test my prototype.

1 Like

Hi @aryanshomray ! For the data, I would like to know what does your architecture look like and what is the amount of data you would need for it. For instance, if you need 1 subject (3D data) to do registration, I would take the data that is already available in DIPY. Look at the tutorials for instance. If you would like to work with a larger set, let me know and I can point you to some good and easy to access data!

Hi @Adit_Kotwal! Thank you for your interest! Yes, do take a look and reach out with any questions you have :slight_smile:

@ShreyasFadnavis. Hey, thanks for all the resources you have provided. I has been great learning through all the suggestion.
I have questions about implementation.

  1. I have been working on architecture for this project idea and among the dataset in DIPY which would be better to use for the initial stage?
  2. and Is this project supposed to be stand-alone that uses the DIPY library OR the deep learning architecture will be integrated into DIPY API in this project?
  3. And is it okay If I contact you regarding deep net architecture and other stuff related to this project through your email?

Thanks again for all your resources

Hi @rahullabs! Sure! Feel free to send me the model :slight_smile:
It should be a part of DIPY (but that is trivial to integrate). That being said, we support only Tensorflow 2.0!

For the dataset, you can use any openly available datasets. See the dataset, I can suggest based on what you propose! OASIS would be a good resource to look at…

1 Like

Hi @ShreyasFadnavis, Thanks alot for sharing the datasite. I have requested the access for the dataset. Additionally, I have created some pull requests, could you please take a look at them?
Thank you

Yes, looks good! an you address my recent comment on your latest PR @KUNAL_MEHTA?

Yes I have merged them into patch-7 and patch-8. The starting three PR’s I have not combined, since these were the past fixes which needed some changes and made those changes there iteself. Shall, I combine them too into one?

Hi @bramsh and @ShreyasFadnavis , I am Dibyakanti , a sophomore at IIT Guwahati. I have worked with tensorflow and have trained generative adversarial networks. I would like to contribute to this project , where should I start?

Hello Dibyakanti @D_Kumar,

There are resources and directions on how to contribute to DIPY mentioned in the comments above on this thread. Please read them.

1 Like

Greetings @ShreyasFadnavis @bramsh
I am Satyam sharma, an AI engineer/Researcher specialized in Deep Learning and Natural language processing and a Data scientist specialized in data analytics and machine learning.

Working on 10+ Real-world projects in different fields of Deep learning including Image segmentation, Image augmentation, object detection, image multi-classification, image multi labeling, OCR, etc of different domains like medical, education, finance, manufacturing. (also contain Data science projects)

In my work, I have also worked with building very large scalable systems that are powered by deep learning maintaining efficiency, performance, and accuracy.

My experience includes everything from using

  • Frameworks like Tensorflow, PyTorch, Keras
  • Pre-trained models and state of art models
  • Building Custom models
  • Implementing models from Research papers
  • Building pipelines
  • communication and leading internal teams
  • collecting, cleaning, analyzing and visualizing data.

Having in-depth knowledge and different skill sets to offer to this project, i would love to be a part of the Team and help it reach new heights.

Any pointers on what should I do next, other than mentioned above,
for e.g regarding the proposal template? @ShreyasFadnavis @bramsh

Thanks for reading such a long text, so patiently,
Satyam sharma.

I am Krishna Katyal I am in my junior year of B tech Computer science and I am from India.
I am into Machine Learning for the past two and a half years.
I have published two research papers until now in the International Conference of Applied Biology,
and have contributed to opensource machine learning libraries sciketlearn,matlotlib and

My work on neural-network-based malaria detection was displayed in pycon India 2019
and I got a scholarship from the R language for the UseR conference

I am skilled in c++, R Python and Python-based deep learning frameworks TensorFlow(1.x and 2.0), Keras and Pytorch.
I would love to contribute to this INCF.
@bramsh @ShreyasFadnavis

Hello @ShreyasFadnavis and @bramsh
I am Shrishti Hore in my Third Year of Electronics Engineering at University of Mumbai. I have worked in various projects related to ML/AI and I would love to contribute as I was intrigued by this project and I have gone through all the previous discussions, documents and resources shared by mentors and contributors.
I have also been reading research papers and been working on DIPY for quite some time.
Looking forward to actively and genuinely contribute to this project during as well as even after GSOC.
Thank you very much.
Shrishti Hore

Hi @ShreyasFadnavis and @bramsh
Is there any specific template that I should follow while creating my proposal?

1 Like

I am graduate student at Indiana University, pursuing computer science with specialization in AI, Machine Learning and Deep Learning. I work as a research assistant at IU and my research focuses on domain of graph neural networks. I have good deep learning background and have done some projects! I also have been learning computer vision(deep learning as well as the traditional techniques) this semester. I am working on 3d semantic segmentation problem currently for a computer vision course.
I am looking forward to contribute to this project!
I hope I am not late to this party!
I will have a look at these ideas, contact the mentors!
Everyone stay safe during these times of pandemic!

@bramsh @ShreyasFadnavis

1 Like