johnseenun/thesis-data · Data Studio · Datasets at Hugging Face

text stringlengths 0 5.79k
Train 203 52 140 395 133
Validation 30 14 21 67 25
Test 32 23 32 85 31

Table 5. WSI-level split according to the lesion type.


6/10
N PB UDH FEA ADH DCIS IC
Train 27 120 56 24 28 40 100
Validation 10 11 9 6 8 9 12
Test 7 16 9 11 12 12 20

Table 6. WSI-level split according to the lesion subtype.


Benign Atypical Malignant Total WSIs Total Patients
Train 1460 1011 1186 3657 106
Validation 135 90 87 312 15
Test 242 162 166 570 30

Table 7. RoI-level split according to the lesion type.


N PB UDH FEA ADH DCIS IC
Train 357 714 389 624 387 665 521
Validation 46 43 46 49 41 40 47
Test 81 79 82 83 79 85 81

Table 8. RoI-level split according to the lesion subtype.


Data organization
The BRACS dataset can be publicly accessed and downloaded via the BRACS website27 . Anyone registering and agreeing with
the terms of use (Creative Commons CC0 license) can freely download it. Once registered, the user can access the FTP server
containing all the data.
The data are organized as follows. The WSIs are stored in the Whole Slide Image Set folder, that includes the train,
validation, and test data. Each data split folder is further partitioned in Benign (BY), Atypical (AT) and Malignant (MT) folders,
each of which includes folders corresponding to lesion subtypes. The WSIs are stored as .svs files. All the files follow the same
naming convention. For instance, the file BRACS_1238.svs refers to the slide ID 1238, whose label is defined by the name of
the folder that contains it. The RoIs are stored in the Region of Interest Set folder, which follows the same structure as the WSI
set. The files are stored in .png format, where the file BRACS_1238_PB_32.png refers to the RoI number 32, extracted from
the WSI named BRACS_1238.svs, and labeled as Pathological Benign. The folder also includes a previous_versions archive
that contains a .zip file with data that have been used in a series of publications during the dataset collection process, e.g., 29–32 .
The WSI annotations are stored in the The Whole Slide Image Annotations folder, which follows the same structure as the WSI
set. It includes annotation files in .qpdata format (based on QuPath17 ) for visualizing the RoIs inside their corresponding WSI.
Finally, a summary file is provided as an .xlsx file, which reports for each WSI, its label, reference set (training/validation/test),
corresponding patient ID and the number of associated RoIs, if any. Figure 3 highlights the folder organization of BRACS.




7/10
Figure 3. The organization of BRACS dataset folders.



8/10
References
1. Madabhushi, A. & Lee, G. Image analysis and machine learning in digital pathology: Challenges and opportunities. Med.
image analysis 33, 170–175 (2016).
2. Tizhoosh, H. R. & Pantanowitz, L. Artificial intelligence and digital pathology: challenges and opportunities. J. pathology
informatics 9 (2018).
3. Srinidhi, C. L., Ciga, O. & Martel, A. L. Deep neural network models for computational histopathology: A survey. Med.
Image Analysis 101813 (2020).
4. de Matos, J., Ataky, S. T. M., de Souza Britto, A., Soares de Oliveira, L. E. & Lameiras Koerich, A. Machine learning
methods for histopathological image analysis: A review. Electronics 10, 562 (2021).
5. Araújo, T. et al. Classification of breast cancer histology images using convolutional neural networks. PloS one 12,
e0177544 (2017).
6. Bardou, D., Zhang, K. & Ahmad, S. M. Classification of breast cancer based on histology images using convolutional
neural networks. Ieee Access 6, 24680–24693 (2018).
7. Sudharshan, P. et al. Multiple instance learning for histopathological breast cancer image classification. Expert. Syst. with
Appl. 117, 103–111 (2019).
8. Duggento, A., Conti, A., Mauriello, A., Guerrisi, M. & Toschi, N. Deep computational pathology in breast cancer. In
Seminars in cancer biology (Elsevier, 2020).
9. Benhammou, Y., Achchab, B., Herrera, F. & Tabik, S. Breakhis based breast cancer automatic diagnosis using deep
learning: Taxonomy, survey and insights. Neurocomputing 375, 9–24 (2020).
10. Sharma, S. & Mehra, R. Conventional machine learning and deep learning approach for multi-classification of breast
cancer histopathology images—a comparative insight. J. digital imaging 33, 632–654 (2020).
11. Chugh, G., Kumar, S. & Singh, N. Survey on machine learning and deep learning applications in breast cancer diagnosis.
Cogn. Comput. 1–20 (2021).
12. Idc: Invasive ductal carcinoma. http://www.andrewjanowczyk.com/use-case-6-invasive-ductal-carcinoma-idc-
segmentation/ (2014).
13. Spanhol, F. A., Oliveira, L. S., Petitjean, C. & Heutte, L. A dataset for breast cancer histopathological image classification.
IEEE Transactions on Biomed. Eng. 63, 1455–1462 (2015).
14. Bejnordi, B. E. et al. Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women
with breast cancer. Jama 318, 2199–2210 (2017).
15. Aresta, G. et al. Bach: Grand challenge on breast cancer histology images. Med. image analysis 56, 122–139 (2019).
16. Veta, M. et al. Predicting breast tumor proliferation from whole-slide images: the tupac16 challenge. Med. Image Analysis
54, 111–121 (2019).
17. Bankhead, P. et al. Qupath: Open source software for digital pathology image analysis. Sci. reports 7, 1–7 (2017).
18. Ingegnoli, A. et al. Flat epithelial atypia and atypical ductal hyperplasia: carcinoma underestimation rate. The breast
journal 16, 55–59 (2010).
19. Breast tumors. WHO classification of tumours series, 5th ed., vol.2. WHO Classif. Tumors Editor. Board. Lyon (France):
Int. Agency for Res. on Cancer 2019 (2010).
20. Bach: The grand challenge on breast cancer histology images. https://iciar2018-challenge.grand-challenge.org/Dataset/
(2018).
21. Roux, L. et al. Mitosis detection in breast cancer histological images an icpr 2012 contest. J. pathology informatics 4
(2013).
22. Bioimaging 2015 breast histology classification challenge. http://www.bioimaging2015.ineb.up.pt/dataset.html (2015).

Train 203 52 140 395 133

Validation 30 14 21 67 25

Test 32 23 32 85 31

Table 5. WSI-level split according to the lesion type.

6/10

N PB UDH FEA ADH DCIS IC

Train 27 120 56 24 28 40 100

Validation 10 11 9 6 8 9 12

Test 7 16 9 11 12 12 20

Table 6. WSI-level split according to the lesion subtype.

Benign Atypical Malignant Total WSIs Total Patients

Train 1460 1011 1186 3657 106

Validation 135 90 87 312 15

Test 242 162 166 570 30

Table 7. RoI-level split according to the lesion type.

N PB UDH FEA ADH DCIS IC

Train 357 714 389 624 387 665 521

Validation 46 43 46 49 41 40 47

Test 81 79 82 83 79 85 81

Table 8. RoI-level split according to the lesion subtype.

Data organization

The BRACS dataset can be publicly accessed and downloaded via the BRACS website27 . Anyone registering and agreeing with

the terms of use (Creative Commons CC0 license) can freely download it. Once registered, the user can access the FTP server

containing all the data.

The data are organized as follows. The WSIs are stored in the Whole Slide Image Set folder, that includes the train,

validation, and test data. Each data split folder is further partitioned in Benign (BY), Atypical (AT) and Malignant (MT) folders,

each of which includes folders corresponding to lesion subtypes. The WSIs are stored as .svs files. All the files follow the same

naming convention. For instance, the file BRACS_1238.svs refers to the slide ID 1238, whose label is defined by the name of

the folder that contains it. The RoIs are stored in the Region of Interest Set folder, which follows the same structure as the WSI

set. The files are stored in .png format, where the file BRACS_1238_PB_32.png refers to the RoI number 32, extracted from

the WSI named BRACS_1238.svs, and labeled as Pathological Benign. The folder also includes a previous_versions archive

that contains a .zip file with data that have been used in a series of publications during the dataset collection process, e.g., 29–32 .

The WSI annotations are stored in the The Whole Slide Image Annotations folder, which follows the same structure as the WSI

set. It includes annotation files in .qpdata format (based on QuPath17 ) for visualizing the RoIs inside their corresponding WSI.

Finally, a summary file is provided as an .xlsx file, which reports for each WSI, its label, reference set (training/validation/test),

corresponding patient ID and the number of associated RoIs, if any. Figure 3 highlights the folder organization of BRACS.

7/10

Figure 3. The organization of BRACS dataset folders.

8/10

References

1. Madabhushi, A. & Lee, G. Image analysis and machine learning in digital pathology: Challenges and opportunities. Med.

image analysis 33, 170–175 (2016).

2. Tizhoosh, H. R. & Pantanowitz, L. Artificial intelligence and digital pathology: challenges and opportunities. J. pathology

informatics 9 (2018).

3. Srinidhi, C. L., Ciga, O. & Martel, A. L. Deep neural network models for computational histopathology: A survey. Med.

Image Analysis 101813 (2020).

4. de Matos, J., Ataky, S. T. M., de Souza Britto, A., Soares de Oliveira, L. E. & Lameiras Koerich, A. Machine learning

methods for histopathological image analysis: A review. Electronics 10, 562 (2021).

5. Araújo, T. et al. Classification of breast cancer histology images using convolutional neural networks. PloS one 12,

e0177544 (2017).

6. Bardou, D., Zhang, K. & Ahmad, S. M. Classification of breast cancer based on histology images using convolutional

neural networks. Ieee Access 6, 24680–24693 (2018).

7. Sudharshan, P. et al. Multiple instance learning for histopathological breast cancer image classification. Expert. Syst. with

Appl. 117, 103–111 (2019).

8. Duggento, A., Conti, A., Mauriello, A., Guerrisi, M. & Toschi, N. Deep computational pathology in breast cancer. In

Seminars in cancer biology (Elsevier, 2020).

9. Benhammou, Y., Achchab, B., Herrera, F. & Tabik, S. Breakhis based breast cancer automatic diagnosis using deep

learning: Taxonomy, survey and insights. Neurocomputing 375, 9–24 (2020).

10. Sharma, S. & Mehra, R. Conventional machine learning and deep learning approach for multi-classification of breast

cancer histopathology images—a comparative insight. J. digital imaging 33, 632–654 (2020).

11. Chugh, G., Kumar, S. & Singh, N. Survey on machine learning and deep learning applications in breast cancer diagnosis.

Cogn. Comput. 1–20 (2021).

12. Idc: Invasive ductal carcinoma. http://www.andrewjanowczyk.com/use-case-6-invasive-ductal-carcinoma-idc-

segmentation/ (2014).

13. Spanhol, F. A., Oliveira, L. S., Petitjean, C. & Heutte, L. A dataset for breast cancer histopathological image classification.

IEEE Transactions on Biomed. Eng. 63, 1455–1462 (2015).

14. Bejnordi, B. E. et al. Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women

with breast cancer. Jama 318, 2199–2210 (2017).

15. Aresta, G. et al. Bach: Grand challenge on breast cancer histology images. Med. image analysis 56, 122–139 (2019).

16. Veta, M. et al. Predicting breast tumor proliferation from whole-slide images: the tupac16 challenge. Med. Image Analysis

54, 111–121 (2019).

17. Bankhead, P. et al. Qupath: Open source software for digital pathology image analysis. Sci. reports 7, 1–7 (2017).

18. Ingegnoli, A. et al. Flat epithelial atypia and atypical ductal hyperplasia: carcinoma underestimation rate. The breast

journal 16, 55–59 (2010).

19. Breast tumors. WHO classification of tumours series, 5th ed., vol.2. WHO Classif. Tumors Editor. Board. Lyon (France):

Int. Agency for Res. on Cancer 2019 (2010).

20. Bach: The grand challenge on breast cancer histology images. https://iciar2018-challenge.grand-challenge.org/Dataset/

(2018).

21. Roux, L. et al. Mitosis detection in breast cancer histological images an icpr 2012 contest. J. pathology informatics 4

(2013).

22. Bioimaging 2015 breast histology classification challenge. http://www.bioimaging2015.ineb.up.pt/dataset.html (2015).