Requirements for MR imaging (As specified in the ACRIN 6657 protocol)

Imaging time points: MRI exams were performed within four weeks prior to starting anthracycline-cyclophosphamide chemotherapy (T1, MRI₁), at least 2 weeks after the first cycle of AC and prior to the second cycle of AC (T2, MRI₂), between anthracycline-cyclophosphamide treatment and taxane therapy if taxane was administered (T3, MRI₃), and after the final chemotherapy treatment and prior to surgery (T4, MRI₄). The study schema is shown in Figure 1

Figure 1. CALGB 150007 and ACRIN 6657 study schema.

Imaging protocol: MR imaging was performed on a 1.5 Tesla field strength scanner using a dedicated breast radiofrequency coil. The image acquisition protocol included a localization scan and T2-weighted sequence followed by a contrast-enhanced T1-weighted series. All imaging was performed unilaterally over the symptomatic breast and in the sagittal orientation. The contrast-enhanced series consisted of a high resolution (≤1mm in-plane spatial resolution) three-dimensional, fat-suppressed, T1-weighted gradient echo sequence with TR≤20 ms, TE = 4.5 ms, flip angle ≤ 45º, 16-18 cm field-of-view, minimum matrix 256×192, 64 slices, slice thickness ≤ 2.5 mm. Scan time length for the T1-weighted sequence was required to be between 4.5 and 5 minutes. The sequence was acquired once before contrast injection and repeated at least twice following injection.

Tumor diameter measurement and volumetric analysis: Tumor longest diameter (LD) was measured by the site radiologist as the greatest extent of disease on baseline MR images, including intervening areas of non-enhancing tissue. The same measurement direction was used on all subsequent MRI exams. The primary predictor variable, functional tumor volume (FTV) was measured from contrast-enhanced images using the signal enhancement ratio (SER) method. Volumetric analysis, including Quality Control assessment, was performed centrally at the Breast Imaging Research Program (BIRP) laboratory at University of California at San Francisco (UCSF).

Detailed information about the DICOM data is available in the DICOM Dictionary.

Further information on these studies can be found at:

• ACRIN 6657 Protocol https://www.acr.org/Research/Clinical-Research/ACRIN-Legacy-Trials
• CALGB 150007 https://clinicaltrials.gov/study/NCT01042379

Imaging Data Transfer History

The processing of the MR image data for ACRIN 6657 consisted of the following steps between image acquisition and the creation of this shared data set on TCIA:

• MRI studies were sent from the study centers to the ACRIN Core Lab either via media (DVD) or the TRIAD program
• Image data were de-identified and centrally archived at the ACRIN Core Lab
• Archived data was sent to the Breast Imaging Research Program (BIRP) at the University of California, San Francisco (UCSF) for volumetric analysis.
• De-identified image data, derived analysis maps and segmentations, and ancillary data files were transferred from UCSF to TCIA for data sharing.

While every effort was made to preserve the integrity of both the original image data and image meta-data (DICOM attributes, public and private), multiple file transfers and strict adherence to HIPPA guidelines for patient confidentiality may have resulted in loss of some data. If any questions arise, or patient PHI is found in any data on this collection, please contact the UCSF Breast Imaging Research Program (BIRP). For scientific or other inquiries about this dataset, please contact TCIA’s Helpdesk.

Curated Data Sets

In addition to the complete set of ACRIN 6657 imaging studies (“Level 0” data), the following curated data sets based on UCSF QC assessment, protocol compliance and data completeness are provided:

• Level 1:    MRI Longest Diameter (LD)
• Level 2a:  SER Volume Dataset for ongoing volumetric analyses (updated 9/17/16)
• Level 2b:  SER Volume Dataset for pCR Analysis (Hylton, et al; Radiology 2012)
• Level 3:    SER Volume Dataset for RFS Analysis (Hylton, et al; Radiology, 2016)

The image data sets are accompanied by Excel files with selected patient clinical and outcome data.

Data subset Descriptions

Level 0: Complete I-SPY 1 / ACRIN 6657 MRI Dataset

This data set is comprised of all HIPAA compliant, DICOM compliant MRI series.

Level 0 image data set consists of 847 on-study MRI studies on 222 subjects in the UCSF image database.
One patient in the image data collection (I-SPY ID 1079) does not appear in the Feb. 2, 2011 I-SPY FINAL LOCKED clinical data dump. So no clinical or outcome data is available for this subject.

Level 1: MRI exams for which longest diameter was measured

This data set is comprised of all studies with MRI measured longest diameter (LD) values reported.
839 MRI studies have LD reported in the I-SPY 1 clinical database, of which 5 studies are not present in either the UCSF or ACRIN image data collections (see Table 1).
Level 1 image data set consists of 834 MRI studies on 219 subjects in the UCSF image database

Level 2a: Good SER Volume Dataset – updated 9/3/14, 9/17/16

This data set is comprised of the patient studies which, following quality reviews in 2014 and 2016, were judged to have sufficiently good image quality and protocol compliance for volumetric DCE SER analysis. Rejection criteria included: incomplete volumetric DCE acquisitions, lack of a 2nd post-contrast acquisition, variability in fat suppression across the image, observed patient motion during the DCE acquisition, significant DCE protocol deviations such as changing scan parameters or image position during DCE acquisition.

Level 2a image data set consists of 706 MR studies on 207 subjects in the UCSF image database. These include 7 studies not included in Level 1 (no MRI LD recorded) as listed in Table 2.

Level 2b: SER Volume Dataset Reported in Hylton et al. (Radiology, 2012)

This data set is comprised of the patient studies analyzed for pCR outcome and reported in the 2012 Radiology paper on ACRIN 6657 pCR results *. This data set is not provided as a shared list, as it is not recommended for use in further analysis. It is described here because it is the data set from which the Level 3 (primary aim analysis) set was derived. Inclusion and exclusion was determined by quality and protocol reviews available at that time. In addition to the exclusion criteria listed for Level 2a, studies done with imaging in the axial plane, in violation of the sagittal orientation specified in the trial imaging protocol, were excluded due to processing limitations of the analysis software. Similarly, bi-lateral sagittal acquisitions (alternating left and right volumetric acquisitions) were excluded.

Level 2b image data set consists of 707 MRI studies on 207 subjects in the UCSF image database.

Tables 3 and 4 show the specific inclusion/exclusion differences between Levels 2a and 2b:

Level 3: Subset of Level 2b used in primary aim analysis, reported in Hylton et al. (Radiology, 2016) *

This data set is comprised of the patient studies analyzed for RFS outcome and reported in the 2015 Radiology paper on ACRIN 6657 survival results (Hylton et al, Radiology *). Table 5 shows the 45 patients excluded from the level 2a cohort for this analysis. Please see the publication for specific information on exclusions of patients from this group.

Level 3 image data set consists of 586 MRI studies on 162 subjects in the UCSF image database. This is also the study cohort used as the Test Phase data in the QIN BMMR Challenge.

This shared data set was provided by David Newitt, PhD and Nola Hylton, PhD from the Breast Imaging Research Program at UCSF, in collaboration with ACRIN, CALGB, the I-SPY TRIAL, and TCIA. Many thanks are due to The ACRIN 6657 trial team, The I-SPY 1 TRIAL team, and all the patients participating in these studies.

Funding sources include NIH grants to UCSF (R01 CA132870 and U01 CA151235), ACRIN (UO1 CA079778 and UO1 CA080098), and CALGB (UO1 CA31964 and UO1 CA33601).

The Collection authors suggest the below will give context to this dataset:

• Hylton, N. M., Blume, J. D., Bernreuter, W. K., Pisano, E. D., Rosen, M. A., Morris, E. A., Weatherall, P. T., Lehman, C. D., Newstead, G. M., Polin, S., Marques, H. S., Esserman, L. J., & Schnall, M. D. (2012). Locally Advanced Breast Cancer: MR Imaging for Prediction of Response to Neoadjuvant Chemotherapy—Results from ACRIN 6657/I-SPY TRIAL Radiology.  https://doi.org/10.1148/radiol.12110748     PMC3359517
• Hylton, N. M., Gatsonis, C. A., Rosen, M. A., Lehman, C. D., Newitt, D. C., Partridge, S. C., Bernreuter, W. K., Pisano, E. D., Morris, E. A., Weatherall, P. T., Polin, S. M., Newstead, G. M., Marques, H. S., Esserman, L. J., & Schnall, M. D. (2016). Neoadjuvant Chemotherapy for Breast Cancer: Functional Tumor Volume by MR Imaging Predicts Recurrence-free Survival—Results from the ACRIN 6657/CALGB 150007 I-SPY 1 TRIAL.   https://doi.org/10.1148/radiol.2015150013  PMC4819899

TCIA maintains a list of publications which leverage our data.

1. Al-Tashi, Q., Saad, M. B., Sheshadri, A., Wu, C. C., Chang, J. Y., Al-Lazikani, B., . . . Wu, J. (2023). SwarmDeepSurv: swarm intelligence advances deep survival network for prognostic radiomics signatures in four solid cancers. Patterns. doi: 10.1016/j.patter.2023.100777.
2.  

   Cattell, R. F., Kang, J. J., Ren, T., Huang, P. B., Muttreja, A., Dacosta, S., . . . Duong, T. Q. (2019). MRI volume changes of axillary lymph nodes as predictor of pathological complete responses to neoadjuvant chemotherapy in breast cancer. Clinical Breast Cancer. doi: 10.1016/j.clbc.2019.06.006.

3.  

   Chitalia, R., Pati, S., Bhalerao, M., Thakur, S. P., Jahani, N., Belenky, V., . . . Bakas, S. (2022). Expert tumor annotations and radiomics for locally advanced breast cancer in DCE-MRI for ACRIN 6657/I-SPY1. Sci Data, 9(1), 440. doi: 10.1038/s41597-022-01555-4.

4.  

   Comes, M. C., Fanizzi, A., Bove, S., Didonna, V., Diotaiuti, S., La Forgia, D., . . . Massafra, R. (2021). Early prediction of neoadjuvant chemotherapy response by exploiting a transfer learning approach on breast DCE-MRIs. Sci Rep, 11(1), 14123. doi: 10.1038/s41598-021-93592-z.

5.  

   Drukker, K., Edwards, A., Papaioannou, J., Giger, M., Hahn, H. K., & Mazurowski, M. A. (2020). Long short-term memory networks predict breast cancer recurrence in analysis of consecutive MRIs acquired during the course of neoadjuvant chemotherapy. Paper presented at the SPIE Medical Imaging, Houston TX USA. doi: 10.1117/12.2549044.

6.  

   Du, R., & Vardhanabhuti, V. (2020, 06-08 July 2020). 3D-RADNet: Extracting labels from DICOM metadata for training general medical domain deep 3D convolution neural networks. Paper presented at the Third Conference on Medical Imaging with Deep Learning (MIDL 2020), Montréal, QC, Canada. Retrieved from https://proceedings.mlr.press/v121/du20a.html.

7.  

   Duanmu, H., Huang, P. B., Brahmavar, S., Lin, S., Ren, T., Kong, J., . . . Duong, T. Q. (2020). Prediction of Pathological Complete Response to Neoadjuvant Chemotherapy in Breast Cancer Using Deep Learning with Integrative Imaging, Molecular and Demographic Data. In Medical Image Computing and Computer Assisted Intervention – MICCAI 2020 (Vol. 12262, pp. 242-252). Lima, Peru: Springer. doi: 10.1007/978-3-030-59713-9_24.

8.  

   Fan, M., Xia, P., Liu, B., Zhang, L., Wang, Y., Gao, X., & Li, L. (2019). Tumour heterogeneity revealed by unsupervised decomposition of dynamic contrast-enhanced magnetic resonance imaging is associated with underlying gene expression patterns and poor survival in breast cancer patients. Breast Cancer Res, 21(1), 112. doi: 10.1186/s13058-019-1199-8.

9.  

   Gierlinger, M., Brandner, D., & Zagar, B. G. (2021, March 17th – 18th, 2021). Vessel extraction from breast MR. Paper presented at the OCM 2021-Optical Characterization of Materials, KARLSRUHE | GERMANY. doi: 10.5445/KSP/1000128686.

10.  

    Jahani, N., Cohen, E., Hsieh, M.-K., Weinstein, S. P., Pantalone, L., Hylton, N., . . . Kontos, D. (2019). Prediction of Treatment Response to Neoadjuvant Chemotherapy for Breast Cancer via Early Changes in Tumor Heterogeneity Captured by DCE-MRI Registration. Sci Rep, 9(1), 12114. doi: 10.1038/s41598-019-48465-x.

11.  

    Kang, J., Li, H., Cattell, R., Talanki, V., Cohen, J. A., Bernstein, C. S., & Duong, T. (2020). The contribution of axillary lymph node volume to recurrence-free survival status in breast cancer patients with sub-stratification by molecular subtypes and pathological complete response. Breast Cancer Research. doi: 10.21203/rs.3.rs-57680/v1.

12.  

    Li, R., & Chen, X. (2022). An efficient interactive multi-label segmentation tool for 2D and 3D medical images using fully connected conditional random field. Comput Methods Programs Biomed, 213, 106534. doi: 10.1016/j.cmpb.2021.106534.

13.  

    Massafra, R., Comes, M. C., Bove, S., Didonna, V., Gatta, G., Giotta, F., . . . Paradiso, A. V. (2022). Robustness Evaluation of a Deep Learning Model on Sagittal and Axial Breast DCE-MRIs to Predict Pathological Complete Response to Neoadjuvant Chemotherapy. Journal of Personalized Medicine, 12(6). doi: 10.3390/jpm12060953.

14.  

    Moyya, P. D., Asaithambi, M., & Ramaniharan, A. K. (2022). Progesterone Receptor Status Analysis in Breast Cancer Patients using DCE- MR Images and Gabor Derived Anisotropy Index. Paper presented at the 2022 IEEE International Symposium on Medical Measurements and Applications (MeMeA), Messina, Italy. doi: 10.1109/MeMeA54994.2022.9856476.

15.  

    Nave, O. (2020). Adding features from the mathematical model of breast cancer to predict the tumour size. International Journal of Computer Mathematics: Computer Systems Theory, 5(3), 159-174. doi: 10.1080/23799927.2020.1792552.

16.  

    Pati, S., Thakur, S. P., Hamamcı, İ. E., Baid, U., Baheti, B., Bhalerao, M., . . . Bakas, S. (2023). GaNDLF: the generally nuanced deep learning framework for scalable end-to-end clinical workflows. Communications Engineering, 2(1). doi: 10.1038/s44172-023-00066-3.

Altmetrics

If you have a manuscript you’d like to add please contact TCIA’s Helpdesk.

Version 1: Updated 2015/06/18