CKA-framework

This repository is an official implementation of the paper "Joint lesion detection and classification of breast ultrasound video via a clinical knowledge-aware framework". (The complete code will be added after the manuscript is published.)

Dataset

We collect and annotate an ultrasound video dataset (298 videos) for lesion detection and classification in ultrasound videos, including 170 benign cases and 128 malignant cases.

Code Usage

Installation

Requirements

• Linux, CUDA>=11.3, GCC>=7.5.0

• Python>=3.8

• PyTorch>=1.11.0, torchvision>=0.12.0 (following instructions here)

• Other requirements

  pip install -r requirements.txt

Dataset preparation

Please organize the dataset as following:

code_root/
└── buv/
      ├── rawframes/
      ├── train.json
      └── val.json

Training

For example, the command for training CKA-framework is as following:

python main.py

The configs in main.py can be changed.

Evaluation

After obtaining the trained CKA-framework, then run following command to evaluate it on the validation set:

python validation.py

Notes

The code of this repository is built on https://github.com/fundamentalvision/Deformable-DETR.