/melanoma-detection

An android application to categorize skin lesions as benign or melanoma based on ABCD features.

Primary LanguageJavaGNU General Public License v3.0GPL-3.0

Melanoma detection on a smartphone

This android application analyzes a lesion image and can classify it as benign or suspicious of melanoma. It is developed using computer vision library OpenCV 3.2 and Java in Android Studio.

Requirments

Environment Setup

  • Download & Install Android Studio

  • Download and extract OpenCV Android Pack 3.2.0 folder and then set OPENCV_ANDROID_SDK environment variable pointing to the downloaded folder location

  • Clone this repository and copy pre-built libraries files

  git clone https://github.com/ukalwa/melanoma-detection
  mkdir melanoma-detection\app\src\main\jniLibs
  REM copy the libraries from opencv to this location
  xcopy /E %OPENCV_ANDROID_SDK%\sdk\native\libs melanoma-detection\app\src\main\jniLibs

  • Now open Android Studio and select Open an existing Android Studio Project

  • Navigate to the melanoma-detection directory and select it

  • When it prompts to install necessary tools, select OK.

  • After all the necessary tools are installed, the project is ready to be tested.

  • Click Build -> Make Project to build the apk. If you get errors at this step, please check the Troubleshooting guide.

  • Before running the apk, check to make sure files opencv_java3.so and libactive_contour.so are in the apk which can be found at <project-root>/<module-root>\build\outputs\apk\debug

  • Now Run the project to install the apk on an android device and install OpenCV_manager when prompted.

Steps involved

The code performs following steps:

  1. Reads in dermoscopic lesion image by taking image from the smartphone camera coupled with a 10x lens or reads an image from the SD card.
  2. Preprocess the image by applying color transformation and filtering
  3. Segment the lesion from the image using active contour model
  4. Extract features (Asymmetry, Border irregularity, Colors, Diameter) from the lesion
  5. Classify the lesion based on the features extracted using an SVM classifier and output the result.
  6. Save the processed images and results

License

This code is GNU GENERAL PUBLIC LICENSED.

Troubleshooting

  • If you get an error like fatal error: 'opencv2/imgproc/imgproc.hpp' file not found, it most likely means that the OPENCV_ADNROID_SDK path is incorrect. You can check this the cmake_build_command.txt at this location <project-root>/<module-root>/.externalNativeBuild/cmake/<build-type>/<ABI>/.

    • To resolve this, restart the computer to get the environment variables reflected in Android Studio.

  • If you get an error after installing OpenCV_manager, uninstall it and install the apk in SDK at %OPENCV_ANDROID_SDK%\apk.

  REM for x86 device
  adb shell %OPENCV_ANDROID_SDK%\apk\OpenCV_3.2.0_Manager_3.20_x86.apk
  • If you get any other error, try cleaning the project and building the project. If it doesn't work, please raise an issue.

Contributing

If you have any suggestions or identified bugs please feel free to post them!

Acknowledgements

  • Active contour segmentation is adapted from ofeli project
  • The computer vision library used for this project is OpenCV