• Design, build, train, and test a deep learning pipeline for feature detection using thoracic CT scans of patients’ hearts
  • Obtain thoracic CT scans of patients
  • Design and implement an annotation tool to annotate crop landmarks for the CT scans
  • Implement a normalization pipeline to normalize the CT scans into 3D arrays
  • Design and implement a CNN that performs feature selection on the normalized arrays

See docs folder for a manual with usage and a tutorial.

GUI tool to annotate DICOM images and batch export.

To run, type python CTImageAnnotationTool.py

See docs folder for a manual with usage and a tutorial.

To Run, type python CNN_main.py **kwargs

See docs folder for a manual with usage and a tutorial.

• About half the final presentation slides (3/8)
• Annotation Tool Manual/Tutorial
• Code
  • AnnotationWidget.py
  • CTImageAnnotationTool.py
  • CTImageAnnotationToolMainWindow.py
  • DataExport.py
  • DataNormalization.py (with Yuan's help)
  • DataReader.py
  • DICOMCrossSectionalImage.py
  • HeartLandmarks.py
  • MathUtil.py
  • SpinnerDialComboWidget.py
  • ViewSliceDataWidget.py
  • ViewSliceWidget.py
  • XYSpinnerComboWidget.py

• About half the final presentation slides (4/8)
• Med3DResNet Manual/Tutorial
• Code
  • CNN_batch_run.py
  • CNN_main.py
  • CNN_ops.py
  • CNN_ResNet.py
  • CNN_utils.py
  • Visualization.py (includes most plots/videos)
  • README.md

• Some of the final presentation slides (1/8)
• Code
  • preprocess.py (some preprocessing images, mainly integrated into Luben's GUI)
  • Docker Container Setup and Build