SLED

Unsupervised Skin Lesion Segmentation via Structural Entropy Minimization on Multi-Scale Superpixel Graphs

Installation

Install the required packages listed in the file requirement.txt. The code has been tested on Python 3.10.6.

Usage

In the root directory of this project:

python main.py [-h] [--dataset DATASET]
               [--slic_nsegments] [--slic_compactness COMPACTNESS]
               [--centroid_thresh CENTROID_THRESH] [--self_tuning_k SELF_TUNING_K]
               [--outlier_detection OUTLIER_DETECTION] [--contamination CONTAMINATION]

example: python main.py --dataset ISIC2016

optional arguments:
  -h                    show this help message and exit
  --dataset DATASET     name of dataset (default: ISIC2016)
  --slic_nsegments      SLIC_NSEGMENTS
                        number of superpixels in SLIC segmentation (default: 400)
  --slic_compactness SLIC_COMPACTNESS    
                        compactness of SLIC superpixel algorithm (default: 10)
  --centroid_thresh CENTROID_THRESH      
                        r for centroid thresh in graph construction (default:0.3)
  --self_tuning_k SELF_TUNING_K          
                        K for local scaling parameters in graph construction (default: 30)
  --outlier_detection OUTLIER_DETECTION  
                        outlier detection method in multi-scale mechanism (default: IFOREST)
  --contamination CONTAMINATION          
                        contamination for outlier detection methods (default: 0.1)