Benchmarks
Prostate
Motivations
Prostate cancer (CaP) has been reported on a worldwide scale to be the second most frequently diagnosed cancer of men accounting for 13.6% (Ferlay et al. (2010)). Statistically, in 2008, the number of new diagnosed cases was estimated to be 899, 000 with no less than 258, 100 deaths (Ferlay et al. (2010)).
Magnetic resonance imaging (MRI) provides imaging techniques allowing to diagnose and localize CaP. The I2CVB provides a multi-parametric MRI dataset to help at the development of computer-aided detection and diagnosis (CAD) system.
Datasets
Overview:
In order to provide a wide dataset, the multi-parametric data have been acquired from two different commercial scanners:
- A 1.5 Tesla General Electric (GE) scanner. To come soon!!!
- A 3.0 Tesla Siemens scanner.
Modalities:
The I2CVB provides a multi-parametric MRI dataset. The modalities available the different dataset are:
- T2-Weighted (T2-W) MRI.
- Dynamic Contrast Enhanced (DCE) MRI.
- Diffusion Weighted Imaging (DWI) MRI.
- Magnetic Resonance Spectroscopic Imaging (MRSI).
Furthermore, the additional Apparent Diffusion Coefficient (ADC) maps are provided for the data acquired with the Siemens scanner.
Image format:
Regarding some technical aspects of the dataset, the T2-W MRI, DCE MRI and DWI MRI, ADC will be delivered in DICOM format.
Regarding the MRSI data will be either provided in RDA (Siemens) or DICOM (GE) format.
Finally all the ground-truth images for each modalities will be provided in DICOM format.
Ground-truth:
For each modality, a set of ground-truth is provided. The ground-truth is composed of four different classes: (i) prostate gland, (ii) peripheral zone (PZ), (iii) central gland (CG), (iv) CaP.
Links:
The datasets are available in Zenodo. Detailed information about each set can be found in:
In addition, the following publications used a subset of data which are available:
- T2-W-MRI, DCE-MRI, ADC, and MRSI from the Siemens 3 T have been used in [2] and the data are available there. Additionally, the associated source is available there.
- T2-W-MRI and DCE-MRI from the Siemens 3 T dataset have been used in [3] and the data are available there. Additionally, the associated source is available there.
Citation:
If you use this dataset for scientific publication, cite the following:
[1] G. Lemaitre, R. Marti, J. Freixenet, J. C. Vilanova, P. M. Walker, and F. Meriaudeau, "Computer-Aided Detection and Diagnosis for prostate cancer based on mono and multi-parametric MRI: A Review", Computer in Biology and Medicine, vol. 60, pp 8 - 31, 2015. [link]
Give us your feedbacks ...
We are still pursuing to add new data. Follow us to know when new data will be available.
Publications
The following list of publications used our dataset. Refer to them regarding a description of the algorithms and a summary of the results. We encourage the authors to provide in some way (Github, ...) the source code.
Your publication does not appear in this webpage, please contact us with all the necessary information.
Related publications:
[2] G. Lemaitre, F. Meriaudeau, J. Freixenet, J. C. Vilanova, and R. Marti "Computer-aided diagnostic for prostate cancer using multi-parametric magnetic resoance imaging", PhD Thesis Submitted. [pdf]
[3] G. Lemaitre, R. Marti, M. Rastgoo, J. Massich, J. Freixenet, A. Meyer-Baese, J. C. Vilanova, and F. Meriaudeau "Automatic prostate cancer detection through DCE-MRI images: all you need is a good normalization", Medical Image Analysis Submitted. [pdf]
[4] R. Trigui, J. Mitéran, P. M. Walker, L. Sellami, and A. B. Hamida, "Automatic classification and localization of prostate cancer using multi-parametric MRI/MRS", Biomedical Signal Processing and Control vol. 31, pp 189 - 198, January 2017. [pdf]
[5] A. Fabijańska, "A novel approach for quantification of time–intensity curves in a DCE-MRI image series with an application to prostate cancer", Computer in Biology and Medicine vol. 73, pp 119 - 130, June 2016. [pdf]
[6] R. Trigui, J. Mitéran, L. Sellami, P. M. Walker, and A. B. Hamida, "A classification approach to prostate cancer localization in 3T multi-parametric MRI", 2nd International Conference on Advanced Technologies for Signal and Image Processing (ATSIP) 2016. Monastir: Tunisia (March 2016). [pdf]
[7] G. Lemaitre, M. Rastgoo, J. Massich, J. C. Vilanova, P. M. Walker, J. Freixenet, A. Meyer-Baese, F. Meriaudeau, and R. Marti, "Normalization of T2W-MRI prostate images using Rician a priori", SPIE Medical Imaging 2016. San Diego: USA (February 2016). [pdf] [source]
[8] G. Lemaitre, J. Massich, R. Marti, J. Freixenet, J. C. Vilanova, P. M. Walker, D. Sidibe, and F. Meriaudeau, "A Boosting Approach for Prostate Cancer Detection using Multi-parametric MRI", International Conference on Quality Control and Artificial Vision (QCAV) 2015. Le Creusot: France (June 2015). [pdf] [source]
Ferlay, J., Shin, H.R., Bray, F., Forman, D., Mathers, C., Parkin, D.M., 2010. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int. J. Cancer 127, 2893–2917.
Retinopathy
Motivations
Diabetic Retinopathy (DR), and more particularly Diabetic Macular Edema (DME), are leading causes of irreversible vision loss and the most common eye diseases in individuals with diabetes.
Taking into account that the number of individuals affected by diabetes dieases are expected to grow exponentially in the next decade (Wild et al. (2014)), developing methodologies for early detection and treatment of DR and DME has become a priority to prevent adverse effects.
Datasets
Publications
The following list of publications used our dataset. Refer to them regarding a description of the algorithms and a summary of the results. We encourage the authors to provide in some way (Github, ...) the source code.
Your publication does not appear in this webpage, please contact us with all the necessary information.
Related publications:
J. Massich, M. Rastgoo, G. Lemaitre, C. Y. Cheung, T. Y. Wong, D. Sidibe, and F. Meriaudeau, "Classifying DME vs Normal SD-OCT volumes: A review", 23rd International Conference on Pattern Recognition (ICPR) 2016. Cancun: Mexico (December 2016). [pdf] [source]
K. Alsaih, G. Lemaitre, J. Massich, M. Rastgoo, D. Sidibe, T. Y. Wong, E. Lamoureux, D. Milea, C. Leung, and F. Meriaudeau, "Classification of SD-OCT volumes with multi-pyramids, LBP and HoG descriptors: Application to DME detection", 38th International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) 2016. Orlando: USA (August 2016). [pdf] [source]
G. Lemaitre, M. Rastgoo, J. Massich, C. Y. Cheung, T. Y. Wong, E. Lamoureux, D. Milea, F. Meriaudeau, and D. Sidibe, "Classification of SD-OCT Volumes using Local Binary Patterns: Experimental Validation for DME detection", Journal of Ophthalmology vol 2016, Mai 2016. [pdf] [source]
G. Lemaitre, M. Rastgoo, J. Massich, S. Sankar, F. Meriaudeau, and D. Sidibe, "Classification of SD-OCT volumes with LBP: Application to DME detection", Ophthalmic Medical Image Analysis Workshop (OMIA), Medical Image Computing and Computer Assisted Interventions (MICCAI) 2015 Munich: Germany (October 2015). [pdf] [source]
Wild, S., Roglic, G., Green, A., Sicree, R., & King, H. (2004). Global prevalence of diabetes estimates for the year 2000 and projections for 2030. Diabetes care, 27(5), 1047-1053.
ICCVB in a nutshell
Vision
Democratize the access to research
The first need in modern research, regardless its application domain, is related to the access to reliable data for its subsequent study.
However, data gathering is an entrance barrier for most of the researchers mainly due to factors as diverse costs, infrastructure, availability, etc. Moreover, isolated endeavours to gather these data without granting public access lead to the creation of muda ("waste"): waste of resources and inability to compare results and validate conclusions.
Despite being highlighted by numerous research works, the lack of usable, public, reliable, and accessible data remains disregarded in many fields. The I2CVB is a wake-up call for addressing and breaking the entrance barriers in research due to data and/or isolation by applying collaborative strategies.
Mission
Open data; evaluation methods; comparison framework; reporting platform
The lack of common data combined with non-aggregated assessing strategies result in non-existent or misleading comparisons make difficult to acknowledge relevant novel methodologies. A common duty to the research and development communities is to overcome these limitations, which can be successfully addressed by co-creation and collaborative work.
I2CVB aims at serving as foundation for collecting and sharing data as well as providing common evaluation methodologies. Furthermore, the use of common data and evaluation is the only way to achieve fair comparison.
Protagonists
Research groups and individuals from all walks of life to shape a transparent community
I2CVB creates for everyone the opportunity to pursue common goals through sharing, collaborating and team-working, to empower the individuals by taking advantages of personal skills and resources. As a consequence, young researchers will find an eco-system for self-improvement in which work will be rewarded through benchmarking compilation.
Strategy
Transfer successful practises from Free Software and Quality Management
I2CVB community challenges the impossible as well as the current status quo in research. Therefore, we strive to settle a multi-skilled community pursuing common goals to achieve excellence through collaborative continuous improvement.
At I2CVB, we believe that Free Software and Quality Management have already reshaped the world and that it is time to apply some of the successful practices learned in such domains to expand the boundaries of research in computer vision and specially for the medical imaging case.
Acknowledgement
Up to now, despite hoping to extend the following list, acknowledges:
I2CVB thanks all the anonymous patients who kindly shared their medical data.
I2CVB thanks the Generalitat de Catalunya, Spanish government, Conseil Regional de Bourgogne, Universitat de Girona for supporting the persons who generate the material shared in I2CVB by means of the following grants: FI-DGR 2012, MOY 2011/018/06, TIN2007-6055, TIN2011-2370, CRB 2013-9201AAO049S02890.
I2CVB thanks the Universitat de Girona (FP7-306088) and GitHub to sustain I2CVB in terms of infrastructure for storing and sharing data.
I2CVB thanks the different collaborators with whom this initiative came into being.
Current contacts
Questions, comments and remarks can be sent to:
Joan Massich Vall
12, Rue de la Fonderie
71200 Le Creusot
P: (+33) 000000000
joan_massich-vall@etu.u-bourgogne.fr
