https://github.com/enkiwang/ps-oct

Polarization-sensitive optical coherence tomography
https://github.com/enkiwang/ps-oct

deep-learning medical-imaging optical-coherence-tomography osteoarthritis-severity-detection ps-oct

Last synced: 2 days ago
JSON representation

Polarization-sensitive optical coherence tomography

• Host: GitHub
• URL: https://github.com/enkiwang/ps-oct
• Owner: enkiwang
• Created: 2022-03-22T09:26:35.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2022-03-22T09:51:24.000Z (over 2 years ago)
• Last Synced: 2023-07-30T14:21:15.910Z (over 1 year ago)
• Topics: deep-learning, medical-imaging, optical-coherence-tomography, osteoarthritis-severity-detection, ps-oct
• Language: Python
• Homepage:
• Size: 39.1 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# PS-OCT

This repository provides codes for "Polarization-sensitive optical coherence tomography with deep learning for detecting the local distribution of osteoarthritis severities". 

## Regression 

In regression case, we experiment on two types of regression labels: coarse labels and dense labels.

For coarse label regression:

```python

GPU_ID=0

python main.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='vgg16' --step_size 20 --gamma 0.2

python main.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='resnet18' --step_size 20 --gamma 0.2

python main.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='densenet121' --step_size 20 --gamma 0.2

python main.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='mobilenetv2' --step_size 20 --gamma 0.2

```

Similarly, you can experiment on "intensity" images by changing "phase" to "intensity".

For dense label regression:

```python

GPU_ID=0

python main_dense.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='vgg16' --step_size 20 --gamma 0.2

python main_dense.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='resnet18' --step_size 20 --gamma 0.2

python main_dense.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='densenet121' --step_size 20 --gamma 0.2

python main_dense.py --img_type='phase' --epoch=150 --gpu_id=${GPU_ID} --model_select='mobilenetv2' --step_size 20 --gamma 0.2

```

## Classification

For PS-OCT classification:

```python

GPU_ID=0

python main_class.py --img_type='phase' --num_class 2 --epoch=150 --gpu_id=${GPU_ID} --model_select='vgg16' --step_size 7 --gamma 0.2

python main_class.py --img_type='phase' --num_class 2 --epoch=150 --gpu_id=${GPU_ID} --model_select='resnet18' --step_size 7 --gamma 0.2

python main_class.py --img_type='phase' --num_class 2 --epoch=150 --gpu_id=${GPU_ID} --model_select='densenet121' --step_size 7 --gamma 0.2

python main_class.py --img_type='phase' --num_class 2 --epoch=150 --gpu_id=${GPU_ID} --model_select='mobilenetv2' --step_size 7 --gamma 0.2

```