Advantages:
- Efficient and Accurate Ground Truth Generation
- Improved Performance in Histopathology Analysis
- Increased Dataset Size for Better Training
Business Summary:
The integrated technology of Bootstrapped Semantic Preprocessing (BSP) sets it apart in the medical histopathology field by enabling effective automatic labeling for medical datasets with limited ground truth. By addressing challenges such as inconsistent imaging conditions and costly ground truth generation, the BSP algorithm transforms unknown samples to align with the distribution of the training set. This advancement greatly enhances the generation of automatic ground truth data using active deep learning, effectively addressing issues associated with inconsistent sample imaging while reducing associated costs. The technology presents a unique and efficient method for accurate histopathology analysis, distinguishing it from conventional approaches in the industry.
Figure 1 – The image depicts the preprocessing algorithm, a fundamental element of methodology. Two components comprise the algorithm: Semantic Preprocessing (SP) and Bootstrap Preprocessing (BSP). SP enhances the semantic information of input images via incremental gamma adjustment. BSP refines segmentation maps using an ensemble and prior maps in an iterative manner. A specialist evaluates the final map for potential incorporation as a new ground truth. This iterative procedure enhances the quality of a map over time.
Desired Partnerships:
- License
- Sponsored Research
- Co-Development