A comparison of algorithms and humans for mitosis detection

We consider the problem of detecting mitotic figures in breast cancer histology slides. We investigate whether the performance of state-of-the-art detection algorithms is comparable to the performance of humans, when they are compared under fair conditions: our test subjects were not previously exposed to the task, and were required to learn their own classification criteria solely by studying the same training set available to algorithms. We designed and implemented a standardized web-based test based on the publicly-available MITOS dataset, and compared results with the performance of the 6 top-scoring algorithms in the ICPR 2012 Mitosis Detection Contest. The problem is presented as a classification task on a balanced dataset. 45 different test subjects produced a total of 3009 classifications. The best individual (accuracy = 0.859 ± 0.012), is outperformed by the most accurate algorithm (accuracy = 0.873 ± 0.004). This suggests that state-of-the-art detection algorithms are likely limited by the size of the training set, rather than by lack of generalization ability.

[1]  [Mammary pathology]. , 1991, Zentralblatt fur Pathologie.

[2]  T. Kuopio,et al.  Mitotic count in breast cancer. The influence of training on performance. , 1996, Pathology, research and practice.

[3]  Joachim M. Buhmann,et al.  Computational Pathology Analysis of Tissue Microarrays Predicts Survival of Renal Clear Cell Carcinoma Patients , 2008, MICCAI.

[4]  A. Madabhushi,et al.  Histopathological Image Analysis: A Review , 2009, IEEE Reviews in Biomedical Engineering.

[5]  Takeo Kanade,et al.  Cell segmentation in microscopy imagery using a bag of local Bayesian classifiers , 2010, 2010 IEEE International Symposium on Biomedical Imaging: From Nano to Macro.

[6]  Luca Fiaschi,et al.  Learning-based mitotic cell detection in histopathological images , 2012, Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012).

[7]  Chao-Hui Huang,et al.  Automated mitosis detection based on eXclusive Independent Component Analysis , 2012, Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012).

[8]  Akira Saito,et al.  Mitotic Figure Recognition: Agreement among Pathologists and Computerized Detector , 2011, Analytical cellular pathology.

[9]  J. Schmidhuber,et al.  Neural Networks for Segmenting Neuronal Structures in EM Stacks , 2012 .

[10]  Johannes Stallkamp,et al.  Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition , 2012, Neural Networks.

[11]  Nasir M. Rajpoot,et al.  A gamma-gaussian mixture model for detection of mitotic cells in breast cancer histopathology images , 2012, Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012).

[12]  Maria S. Kulikova,et al.  Mitosis detection in breast cancer histological images An ICPR 2012 contest , 2013, Journal of pathology informatics.

[13]  Luca Maria Gambardella,et al.  Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks , 2013, MICCAI.

[14]  Luca Maria Gambardella,et al.  Fast image scanning with deep max-pooling convolutional neural networks , 2013, 2013 IEEE International Conference on Image Processing.

[15]  Mitko Veta,et al.  Detecting mitotic figures in breast cancer histopathology images , 2013, Medical Imaging.

[16]  H. Irshad Automated mitosis detection in histopathology using morphological and multi-channel statistics features , 2013, Journal of pathology informatics.

[17]  Eric Cosatto,et al.  Classification of mitotic figures with convolutional neural networks and seeded blob features , 2013, Journal of pathology informatics.