LIVERSCANPLUSPLUS
Description
The LIVERSCANPLUSPLUS - v17 dataset addresses this need by providing a standardized, annotated, and augmented collection of liver lesion images. Created an end-to-end computer vision platform, this dataset is designed to facilitate research in medical imaging and computer vision. It is publicly available and includes detailed annotations, pre-processing, and augmentation to support the development of accurate and generalizable models. The LIVERSCANPLUSPLUS - v17 dataset consists of 822 images annotated for liver lesion detection. The dataset is divided into three subsets for training, validation, and testing, as shown in Table 1. Table 1: Dataset Splits Subset Percentage Number of Images Purpose Training Set 87% 717 Model training and parameter tuning Validation Set 9% 70 Hyperparameter optimization Test Set 4% 35 Final model evaluation
Files
Institutions
- Minia University Faculty of Science