Skip to content

Latest commit

 

History

History
100 lines (71 loc) · 4.09 KB

Osteosarcoma.md

File metadata and controls

100 lines (71 loc) · 4.09 KB

Osteosarcoma Tumor Assessment

Dataset Information

The Osteosarcoma-Tumor-Assessment dataset uses archival samples from 50 patients who were treated at Children's Medical Center Dallas between 1995 and 2015. Four patients (out of the 50) were selected by pathologists based on the diversity of their tumor specimens after surgical resection. Each image is labeled as non-tumor, viable tumor, or necrotic based on the predominant cancer type in the image. Annotations were made by two medical experts. All images were assigned to two pathologists for the annotation task, but each image has only one annotation, as any given image was annotated by a single pathologist. The dataset consists of 1,144 images with a size of 1024 x 1024 at 10X magnification, distributed as follows: 536 (47%) non-tumor images, 263 (23%) necrotic tumor images, and 345 (30%) viable tumor images.

Osteosarcoma is the most common type of bone cancer, primarily affecting adolescents aged 10 to 14 years. This dataset comprises histological images of osteosarcoma stained with Hematoxylin and Eosin (H&E). The data was collected by a team of clinical scientists at the University of Texas Southwestern Medical Center in Dallas.

Dataset Meta Information

Dimensions Modality Task Type Anatomical Structures Number of Categories Data Volume File Format
2D pathology Classification Bone 3 1144 PNG

Resolution Details

Dataset Statistics size
min (1024, 1024)
median (1024, 1024)
max (1024, 1024)

Label Information Statistics

Metric Tumor
0 (Non Tumor) 415
1 (Necrotic Tumor) 267
2 (Viable Tumor) 233

Visualization

Paper Visualization.

File Structure

The dataset file structure is as follows:

Osteosarcoma-Tumor-Assessment
├── image
│   ├── train
│   │   ├── xxx.jpg
│   │   ├── xxx.jpg
│   │   ├── xxx.jpg
│   │   ├── xxx.jpg
│   │   └── ...
│   └── val
│   │   ├── xxx.jpg
│   │   ├── xxx.jpg
│   │   ├── xxx.jpg
│   │   ├── xxx.jpg
│   │   └── ...
    ├── train.txt
    └── val.txt

Authors and Institutions

Rashika Mishra (Department of Computer Science, University of Texas at Dallas)

Ovidiu Daescu (Department of Computer Science, University of Texas at Dallas)

Patrick Leavey (University of Texas Southwestern Medical Center)

Dinesh Rakheja (University of Texas Southwestern Medical Center)

Source Information

Official Website: https://www.cancerimagingarchive.net/collection/osteosarcoma-tumor-assessment/

Download Link: https://www.cancerimagingarchive.net/collection/osteosarcoma-tumor-assessment/

Article Address: https://doi.org/10.1007/978-3-319-59575-7_2

Publication Date: 2019

Citation

@inproceedings{mishra2017histopathological,
  title={Histopathological diagnosis for viable and non-viable tumor prediction for osteosarcoma using convolutional neural network},
  author={Mishra, Rashika and Daescu, Ovidiu and Leavey, Patrick and Rakheja, Dinesh and Sengupta, Anita},
  booktitle={Bioinformatics Research and Applications: 13th International Symposium, ISBRA 2017, Honolulu, HI, USA, May 29--June 2, 2017, Proceedings 13},
  pages={12--23},
  year={2017},
  organization={Springer}
}

Original introduction article is here.