uw-biomedical-ml/irf-segmenter

Deep learning based, intraretinal fluid segmenter for macular OCTs

irf-segmenter

Fully-automated, deep learning based, intraretinal fluid segmenter for optical coherence (OCT) images of the macula

Usage

usage: run.py [-h] [--mode MODE] input_file output_file

positional arguments:
  input_file   Input PNG file
  output_file  Output PNG file

optional arguments:
  -h, --help   show this help message and exit
  --mode MODE  Output mode, 'mask_blend' (default) for masked heatmap output,
               'mask' for binary mask output, 'blend' for blended heatmap

Example

Input image:

Mask Blended image:

Blended image:

Binary Mask image:

Requirements

You must install the following Python packages:

• TensorFlow (0.12.1)
  • See installation instructions.
• Keras (1.2.1)
  • See installation instructions.
• HDF5 and h5py
• Pillow
• Matplotlib
• cv2

conda install --file requirements.txt

or

pip install -r requirements.txt

If having troubles with import cv2, try:

conda install --channel https://conda.anaconda.org/menpo opencv3

Installation

git clone https://github.com/uw-biomedical-ml/irf-segmenter