Tumor Subclone Detection From Digital Pathology Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current genomic profiling of tumors often profiles only a single sub-clone, leading to inaccurate information due to tumor heterogeneity, where multiple sub-clones may be intermingled, and existing methods for identifying tumor sub-clones are expensive, time-consuming, and error-prone.
Innovation Solution
An AI-based method for detecting tumor sub-clones using digital pathology images, employing machine learning to identify genetically distinct tumors and their subclonal relationships, enabling targeted spatial profiling and treatment recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If genomic profiling is performed on a single sub-clone, then the profiling process is simplified, but the diagnostic accuracy deteriorates due to tumor heterogeneity
Solution Approach 1:
The patent segments the tumor into multiple sub-clones by identifying distinct neoplastic regions with different morphological features. The system divides the tumor tissue into separate analyzable units, each representing a potential sub-clone, thereby capturing tumor heterogeneity without requiring complex multi-clone profiling procedures
Solution Approach 2:
The system applies local quality analysis by examining specific morphological features (nuclear size, shape, chromatin pattern, nucleoli) in different regions of the tumor. Each region is analyzed with attention to its unique characteristics, allowing differentiation of sub-clones based on local morphological variations rather than requiring comprehensive genomic analysis of all cells
2Ease of operation
If multiple sub-clones are intermingled in a single sample, then the sampling process is simplified, but the measurement accuracy deteriorates
Solution Approach 1:
The system segments intermingled neoplastic cells into distinct clusters based on morphological similarity. By applying unsupervised clustering algorithms to nuclear features, the system separates mixed sub-clones into discrete groups, enabling accurate sub-clone identification even when cells are physically intermingled in the tissue sample
Solution Approach 2:
The system creates morphological feature vectors (copies of key characteristics) for each neoplastic cell and uses these digital representations for clustering analysis. This allows accurate sub-clone differentiation based on replicated morphological data rather than requiring physical separation of cells
3Reliability
If traditional methods are used to identify tumor sub-clones, then the identification process can be performed, but the time consumption and error rate increase
Solution Approach 1:
The patent replaces manual pathological examination and traditional genomic profiling with automated machine learning-based morphological analysis. The system uses computer vision algorithms to automatically extract and analyze nuclear features, perform clustering, and identify sub-clones, eliminating time-consuming manual processes while improving consistency and reducing human error
Solution Approach 2:
The system changes the analysis parameters from comprehensive genomic sequencing to focused morphological feature extraction (nuclear size, shape, chromatin texture, nucleoli characteristics). This parameter transformation enables rapid sub-clone identification using standard pathology images, significantly reducing analysis time while maintaining reliability through multiple feature dimensions
Data Source
Figure 1A
Figure 1B
Figure 2A
AI summary
A computer-implemented method for detecting tumor subclones may include receiving one or more digital images into a digital storage device, the one or more digital images including images of a tumor of a patient, detecting one or more neoplasms in the one or more received digital images for each patient, determining a score for each detected one or more neoplasms, determining a hierarchy dendrogram based on the detected one or more neoplasms and the determined score for each of the detected one or more neoplasms, determining one or more leaf nodes based on the determined hierarchy dendrogram, and determining, based on the determined hierarchy dendrogram, whether there are two or more neoplasms among the detected one or more neoplasms that originated independently.