curated_breast_imaging_ddsm

  • 설명 :

CBIS-DDSM(DDSM의 선별된 유방 영상 하위 집합)은 유방조영술 디지털 데이터베이스(DDSM)의 업데이트되고 표준화된 버전입니다. DDSM은 2,620개의 스캔된 필름 유방조영술 연구의 데이터베이스입니다. 검증된 병리정보와 함께 정상, 양성, 악성 사례를 담고 있습니다.

기본 구성은 전통적인 이미지 분류 설정에서 해결해야 할 작업을 구성하기 위해 ( http://arxiv.org/abs/1708.09427 )의 설명에 따라 원본 유방 조영술에서 추출한 패치로 구성됩니다.

데이터 세트에 포함된 이미지를 다운로드하고 읽으려면 특수 소프트웨어와 라이브러리가 필요하기 때문에 TFDS는 사용자가 원본 DCIM 파일을 다운로드하여 PNG로 변환했다고 가정합니다.

재현 가능한 결과를 보장하려면 PNG 파일을 생성하는 데 다음 명령(또는 이에 상응하는 명령)을 사용해야 합니다.

find $DATASET_DCIM_DIR -name '*.dcm' | \
xargs -n1 -P8 -I{} bash -c 'f={}; dcmj2pnm $f | convert - ${f/.dcm/.png}'

결과 이미지는 <manual_dir>/Mass-Training_P_01981_RIGHT_MLO_1/1.3.6.../000000.png 와 같이 manual_dir 에 넣어야 합니다.

@misc{CBIS_DDSM_Citation,
  doi = {10.7937/k9/tcia.2016.7o02s9cy},
  url = {https://wiki.cancerimagingarchive.net/x/lZNXAQ},
  author = {Sawyer-Lee,  Rebecca and Gimenez,  Francisco and Hoogi,  Assaf and Rubin,  Daniel},
  title = {Curated Breast Imaging Subset of DDSM},
  publisher = {The Cancer Imaging Archive},
  year = {2016},
}
@article{TCIA_Citation,
  author = {
    K. Clark and B. Vendt and K. Smith and J. Freymann and J. Kirby and
    P. Koppel and S. Moore and S. Phillips and D. Maffitt and M. Pringle and
    L. Tarbox and F. Prior
  },
  title = { {The Cancer Imaging Archive (TCIA): Maintaining and Operating a
  Public Information Repository} },
  journal = {Journal of Digital Imaging},
  volume = {26},
  month = {December},
  year = {2013},
  pages = {1045-1057},
}
@article{DBLP:journals/corr/abs-1708-09427,
  author    = {Li Shen},
  title     = {End-to-end Training for Whole Image Breast Cancer Diagnosis using
               An All Convolutional Design},
  journal   = {CoRR},
  volume    = {abs/1708.09427},
  year      = {2017},
  url       = {http://arxiv.org/abs/1708.09427},
  archivePrefix = {arXiv},
  eprint    = {1708.09427},
  timestamp = {Mon, 13 Aug 2018 16:48:35 +0200},
  biburl    = {https://dblp.org/rec/bib/journals/corr/abs-1708-09427},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

curated_breast_imaging_ddsm/patches(기본 구성)

  • 구성 설명 : 석회화 및 대량 사례를 모두 포함하고 이상이 없는 경로를 포함하는 패치입니다. 전통적인 5등급 분류 작업으로 설계되었습니다.

  • 다운로드 크기 : 2.01 MiB

  • 데이터세트 크기 : 801.46 MiB

  • 분할 :

나뉘다
'test' 9,770
'train' 49,780
'validation' 5,580
  • 기능 구조 :
FeaturesDict({
    'id': Text(shape=(), dtype=string),
    'image': Image(shape=(None, None, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=5),
})
  • 기능 문서 :
특징 수업 모양 Dtype 설명
특징Dict
ID 텍스트
영상 영상 (없음, 없음, 1) uint8
상표 클래스 라벨 정수64

심상

curated_breast_imaging_ddsm/original-calc

  • 구성 설명 : 무손실 PNG로 압축된 석회화 사례의 원본 이미지입니다.

  • 다운로드 크기 : 1.06 MiB

  • 데이터세트 크기 : 4.42 GiB

  • 분할 :

나뉘다
'test' 284
'train' 1,227
  • 기능 구조 :
FeaturesDict({
    'abnormalities': Sequence({
        'assessment': ClassLabel(shape=(), dtype=int64, num_classes=6),
        'calc_distribution': ClassLabel(shape=(), dtype=int64, num_classes=10),
        'calc_type': ClassLabel(shape=(), dtype=int64, num_classes=48),
        'id': int32,
        'mask': Image(shape=(None, None, 1), dtype=uint8),
        'pathology': ClassLabel(shape=(), dtype=int64, num_classes=3),
        'subtlety': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
    'breast': ClassLabel(shape=(), dtype=int64, num_classes=2),
    'id': Text(shape=(), dtype=string),
    'image': Image(shape=(None, None, 1), dtype=uint8),
    'patient': Text(shape=(), dtype=string),
    'view': ClassLabel(shape=(), dtype=int64, num_classes=2),
})
  • 기능 문서 :
특징 수업 모양 Dtype 설명
특징Dict
이상 순서
이상/평가 클래스 라벨 정수64
이상/계산_분포 클래스 라벨 정수64
이상/calc_type 클래스 라벨 정수64
이상/ID 텐서 정수32
이상/마스크 영상 (없음, 없음, 1) uint8
이상/병리 클래스 라벨 정수64
이상/미묘함 클래스 라벨 정수64
가슴 클래스 라벨 정수64
ID 텍스트
영상 영상 (없음, 없음, 1) uint8
인내심 있는 텍스트
보다 클래스 라벨 정수64

심상

curated_breast_imaging_ddsm/original-mass

  • 구성 설명 : 무손실 PNG로 압축된 대량 케이스의 원본 이미지입니다.

  • 다운로드 크기 : 966.57 KiB

  • 데이터세트 크기 : 4.80 GiB

  • 분할 :

나뉘다
'test' 348
'train' 1,166
  • 기능 구조 :
FeaturesDict({
    'abnormalities': Sequence({
        'assessment': ClassLabel(shape=(), dtype=int64, num_classes=6),
        'id': int32,
        'mask': Image(shape=(None, None, 1), dtype=uint8),
        'mass_margins': ClassLabel(shape=(), dtype=int64, num_classes=20),
        'mass_shape': ClassLabel(shape=(), dtype=int64, num_classes=21),
        'pathology': ClassLabel(shape=(), dtype=int64, num_classes=3),
        'subtlety': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
    'breast': ClassLabel(shape=(), dtype=int64, num_classes=2),
    'id': Text(shape=(), dtype=string),
    'image': Image(shape=(None, None, 1), dtype=uint8),
    'patient': Text(shape=(), dtype=string),
    'view': ClassLabel(shape=(), dtype=int64, num_classes=2),
})
  • 기능 문서 :
특징 수업 모양 Dtype 설명
특징Dict
이상 순서
이상/평가 클래스 라벨 정수64
이상/ID 텐서 정수32
이상/마스크 영상 (없음, 없음, 1) uint8
이상/질량_마진 클래스 라벨 정수64
이상/질량_형상 클래스 라벨 정수64
이상/병리 클래스 라벨 정수64
이상/미묘함 클래스 라벨 정수64
가슴 클래스 라벨 정수64
ID 텍스트
영상 영상 (없음, 없음, 1) uint8
인내심 있는 텍스트
보다 클래스 라벨 정수64

심상