Downloads: 1
Switch to unified view
| a/README.md | b/README.md | ||
|---|---|---|---|
| 1 | <h1 align="center"> |
1 | <h1 align="center">
|
| 2 | <p align="center">SpineAI Paper with Code</p> |
2 | <p align="center">SpineAI Paper with Code</p> |
| 3 | <img src="imgs/spineAI-logo.png" alt="SpineAI-logo" height="150"> |
3 | </h1> |
| 4 | </h1> |
4 | |
| 5 | 5 | ||
| 6 | |||
| 7 | ## 📄 About |
6 | ## 📄 About |
| 8 | 7 | ||
| 9 | This repository contains **code for our paper published in Radiology Journal (RSNA)**: |
8 | This repository contains **code for our paper published in Radiology Journal (RSNA)**: |
| 10 | 9 | ||
| 11 | > **"Deep Learning Model for Automated Detection and Classification of Central Canal, Lateral Recess, and Neural Foraminal Stenosis at Lumbar Spine MRI"** |
10 | **"Deep Learning Model for Automated Detection and Classification of Central Canal, Lateral Recess, and Neural Foraminal Stenosis at Lumbar Spine MRI"** |
| 12 | 11 | ||
| 13 |  |
12 |  |
| 14 | 13 | ||
| 15 | [Article link](https://pubs.rsna.org/doi/abs/10.1148/radiol.2021204289) _(Published online: May 11 2021)_ |
14 | [Article link](https://pubs.rsna.org/doi/abs/10.1148/radiol.2021204289) _(Published online: May 11 2021)_ |
| 16 | 15 | ||
| 17 | In the paper, we develop and apply AI techniques to automatically detect and classify lumbar spinal stenosis on MRI images. |
16 | In the paper, we develop and apply AI techniques to automatically detect and classify lumbar spinal stenosis on MRI images. |
| 18 | 17 | ||
| 19 | ## 🎓 What’s In This Repo |
18 | ## 🎓 What’s In This Repo |
| 20 | 19 | ||
| 21 | The setups and implementations of our system are documented in their respective folders: |
20 | The setups and implementations of our system are documented in their respective folders: |
| 22 | 21 | ||
| 23 | - [**Object Detection**](Object-Detection/) |
22 | - [**Object Detection**](Object-Detection/) |
| 24 | 23 | ||
| 25 | Based on Tensorflow object detection API, we pick Faster R-CNN with Resnet101 architecture pre-trained on COCO dataset to detect region of interest (ROI). |
24 | Based on Tensorflow object detection API, we pick Faster R-CNN with Resnet101 architecture pre-trained on COCO dataset to detect region of interest (ROI). |
| 26 | 25 | ||
| 27 | - [**Classification**](Classification/) |
26 | - [**Classification**](Classification/) |
| 28 | 27 | ||
| 29 | CNN architecture consisting of six convolutional layers, outputing four-grade classification predictions. |
28 | CNN architecture consisting of six convolutional layers, outputing four-grade classification predictions. |
| 30 | 29 | ||
| 31 | - [**Inference**](Inference/) |
30 | - [**Inference**](Inference/) |
| 32 | 31 | ||
| 33 | Predict relevant spinal regions (ROI) and infer the disease grades, automatic generation of XML outputs and bounding boxes with probability overlays. |
32 | Predict relevant spinal regions (ROI) and infer the disease grades, automatic generation of XML outputs and bounding boxes with probability overlays. |
| 34 | 33 | ||
| 35 | - [**Interpretability**](Interpretability/) |
34 | - [**Interpretability**](Interpretability/) |
| 36 | 35 | ||
| 37 | Explainable AI technique using Integrated Gradients provided by Tensorflow Core. |
36 | Explainable AI technique using Integrated Gradients provided by Tensorflow Core. |
| 38 | 37 | ||
| 39 | 38 | ||
| 40 | ## 🤝 Referencing and Citing SpineAI |
39 | ## 🤝 Referencing and Citing SpineAI |
| 41 | 40 | ||
| 42 | If you find our work useful in your research and would like to cite our Radiology paper, please use the following citation: |
41 | If you find our work useful in your research and would like to cite our Radiology paper, please use the following citation: |
| 43 | 42 | ||
| 44 | ``` |
43 | ```
|
| 45 | @article{hallinan2021deep, |
44 | @article{hallinan2021deep,
|
| 46 | title={Deep learning model for automated detection and classification of central canal, lateral recess, and neural foraminal stenosis at lumbar spine MRI}, |
45 | title={Deep learning model for automated detection and classification of central canal, lateral recess, and neural foraminal stenosis at lumbar spine MRI},
|
| 47 | author={Hallinan, James Thomas Patrick Decourcy and Zhu, Lei and Yang, Kaiyuan and Makmur, Andrew and Algazwi, Diyaa Abdul Rauf and Thian, Yee Liang and Lau, Samuel and Choo, Yun Song and Eide, Sterling Ellis and Yap, Qai Ven and others}, |
46 | author={Hallinan, James Thomas Patrick Decourcy and Zhu, Lei and Yang, Kaiyuan and Makmur, Andrew and Algazwi, Diyaa Abdul Rauf and Thian, Yee Liang and Lau, Samuel and Choo, Yun Song and Eide, Sterling Ellis and Yap, Qai Ven and others},
|
| 48 | journal={Radiology}, |
47 | journal={Radiology},
|
| 49 | pages={204289}, |
48 | pages={204289},
|
| 50 | year={2021}, |
49 | year={2021},
|
| 51 | publisher={Radiological Society of North America} |
50 | publisher={Radiological Society of North America}
|
| 52 | } |
51 | }
|
| 53 | ``` |
52 | ``` |
| 54 | 53 | ||
| 55 | ## :mailbox: Contact |
54 | ## :mailbox: Contact |
| 56 | 55 | ||
| 57 | Address correspondence to J.T.P.D.H. (e-mail: james_hallinan AT nuhs.edu.sg) |
56 | Address correspondence to J.T.P.D.H. (e-mail: james_hallinan AT nuhs.edu.sg) |
| 58 | 57 | ||
| 59 | ### _Disclaimer_ |
58 | ### _Disclaimer_ |
| 60 | 59 | ||
| 61 | _This code base is for research purposes and no warranty is provided. We are not responsible for any medical usage of our code._ |
60 | _This code base is for research purposes and no warranty is provided. We are not responsible for any medical usage of our code._
|