Locality Sensitive Deep Learning for Detection and Classification of Nuclei in Routine Colon Cancer Histology Images
Problem
Detecting and classifying cell nuclei in routine colon cancer histology images, which is difficult due to cellular heterogeneity.
Why it matters
Reliable nuclei detection and classification underpins cancer diagnosis, grading and research.
What the AI did
A Spatially Constrained Convolutional Neural Network (SC-CNN) regresses the likelihood of pixels being nucleus centers for detection, and a Neighboring Ensemble Predictor (NEP) coupled with a CNN classifies detected nuclei without requiring segmentation.
Results
Evaluated on a large dataset of colorectal adenocarcinoma images with more than 20,000 annotations; performance metrics not stated in source (abstract truncated).
Impact
Widely cited (1217 citations) peer-reviewed deep learning method for nuclei analysis in cancer histology.
Record
| Venue | IEEE Transactions on Medical Imaging |
|---|---|
| Published | 2016 |
| Work type | article |
| Authors | Korsuk Sirinukunwattana, Shan E Ahmed Raza, Yee‐Wah Tsang, David R. J. Snead, Ian A. Cree, Nasir Rajpoot |
| Citations | 1,217 (OpenAlex, at import time) |
| References | 51 |
| Model or method | SC-CNN and NEP-coupled CNN |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1109/tmi.2016.2525803 |
| Open access | closed |
| Keywords | deep learning, histopathology, nuclei detection, colon cancer, convolutional neural network |
Importance rating
real but narrower advance
Nuclei detection/classification in colon cancer histology supports diagnosis and research; abstract notes prospective pathology benefit but gives no scale—a narrower methodological advance.