
Locality Sensitive Deep Learning for Detection and Classification of Nuclei in Routine Colon Cancer Histology Images
Abstract of Locality Sensitive Deep Learning for Detection
The proposed approaches for detection and classification do not require segmentation of nuclei.
We have evaluated them on a large dataset of colorectal adenocarcinoma images, consisting of more than 20,000 annotated nuclei belonging to four different classes.
There are several factors that hinder automatic approaches for detection and classification of cell nuclei.
On the other hand, complex tissue architecture clutter of nuclei, and diversity of nuclear morphology pose a challenging problem.
Variability in the appearance of the same type of nuclei, both within
and across different sample, is another factor that makes classification of individual nucleus equivalently difficult.In this paper, we present novel locality sensitive deep learning approaches to detect
The use of small patches not only increases the amount of training data which is crucial for CNNs, but essentially also localizes our analysis to small nuclei in images.
Conclusion
Locality Sensitive Deep Learning for Detection.In this study, we presented deep learning approaches sensitive to the local neighborhood for nucleus detection and classification in routine stained histology images of colorectal adenocarcinomas.
The evaluation was conducted on a large dataset with more than 20,000 annotated nuclei from samples of different histologic grades.