Histopathology Classification of Colorectal Cancer Whole Slide Images Using Color Features with Deep Residual Transfer Learning
Bibliographic record
Abstract
Colorectal cancer (CRC) is an emerging global health concern. An average of 73 Canadians will be diagnosed with CRC every day and 27 Canadians will lose their life as a result of it. CRC accounts for 12% of all cancer deaths in Canada in the year 2020. Early and accurate diagnosis is vital in saving lives as it significantly influences the length of survival of the patient. Deep learning can be leveraged to aid in the task of identifying cancerous cells within pre-cancerous tissue samples, which are taken from colorectal polyps of patients for CRC screening. In this study, an attempt to improve existing supervised methods of classification of colorectal cancer is made. By revamping/improving the deep learning architecture in ResNet. The network will be trained on a much larger and relevant dataset of colorectal WSI (Whole Slide Image) patches. This study aims to attain better overall accuracy by incorporating color features, which have not been concentrated on in previous studies. All while retaining similar performance as compared to existing state-of-the-art methods of CRC classification. Four network models are applied to a large histopathological dataset. All network models are variations of Residual networks at multiple depths. The best results are attained using a pre-trained ResNet-50 model. The overall results show that the residual network performs similarly to the much deeper DenseNet-121 model and better than the cell level framework described in a previous study. The ResNet-50 model achieved 88.58%, 92.04%, 81.86%, 86.65% Accuracy, Precision, Recall and F1-Score respectively.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".