ColoPola dataset is Colorectal cancer Polarimetric images dataset

Polarimetric imaging is increasingly used in biological research, such as tissue characterization and cancer detection. To support classification models, this study introduces ColoPola, a publicly available colorectal cancer polarimetric image dataset (288 healthy, 284 malignant), each with 36 images for different polarization states - totaling 20,592 images. Five classification models (CNN, CNN_2, EfficientFormerV2, DenseNet, EfficientNetV2) were tested; EfficientNetV2 performed best, achieving F1 scores above 0.95 on validation and testing sets. ColoPola shows strong potential for clinical diagnostic applications in colorectal cancer.