Medical Image Registration Using Deep Learning

Deep learning based 3D Medical Image registration
————-Introduction–Section 1————— 2 Pages
1.1. Convolutional neural network
1.2. Autoencoder
1.3. Recurrent neural network
1.4. Reinforcement learning
1.5. Generative adversarial network
————–Literature and Related work ——Section 2————— 6 Pages
Overview of previous work in 3D Medical Image Registration using Deep learning Categorized
into 3 category:
2.1 Deep Iterative Registration
2.1.1 Deep Similarity Based Registration
2.1.2 Reinforcement Learning-based Registration
2.2 Supervised Registration
2.2.1 Fully Supervised Registration
2.2.2 Weakly Supervised Registration
2.2.3 Dual Supervised Registration
2.3 Unsupervised Registration
2.3.1 Similarity Metric-based Registration
2.3.2 GAN-Based Registration
————–Section 3————— 2 Pages
Explain loss functions used in the following papers by writing formula and explaining each term in your own words
3.1 Voxemorph (https://arxiv.org/pdf/1802.02604.pdf)
3.2 LapIRN (https://arxiv.org/pdf/2006.16148.pdf)
3.3 CLapIRN (https://arxiv.org/pdf/2106.12673.pdf)
Instructions:
1. In section 1 the subsections (1.1 – 1.5) should give general idea about the topic.
Each subsection is Type of neural network architecure. It should be refered from different
research papers. And must be written in Data Sciene and Machine learning architecutre context.
2. All the work must not be plagrized.
3. It should not be a paraphrased copy of the reference paper. None of the content should be a paraphrased version of abstract of the reference paper
4. All subsections (1.1 – 2.3) should have atleast 5 – 6 reference papers.
Almost all sentecnce in a paragraph should be written by referencing a different research paper and should be on your own words since it is related work
section.
i.e One paragraph or sub section should not be completly written from one reference paper.
5. In section 2 the subsections(2.1 – 2.3) should be written in terms of Medical Image
Registration. Attached papers will give overview of how it should be written.
6. The paper should be written from Computer Science point of view, therfore the focus should be on Medical Image
Registration algorithms using Deep learning. It should not focus on Doctors, lab-technician, medical expert
and other stake holders point of view.
7. Section 1 should be written for 2 pages (single-spaced, without diagrams and tables)
6. Section 2 Should be written for 6 pages (single-spaced, without diagrams and tables)
8. Section 3 Should be written for 2 pages (single-spaced) with equations provided in reference papers