Machine Learning for Tumor Shape Classification
Focus: Leveraging Machine Learning (ML) to classify tumor shapes and analyze their complexity from large datasets of tumor point sets.
Key Contributions:
Developed custom shape descriptors, such as the "bellyshape" function, to quantify cell area and internal shape complexity by fitting non-overlapping circles within tumor contours.
Utilized Principal Component Analysis (PCA) to reduce dimensionality and identify key features for classification.
Trained supervised ML models, such as Support Vector Machines (SVM) and Decision Trees, to categorize tumor shapes into distinct classes, improving classification accuracy and scalability.
Addressed challenges in dataset variability by implementing data augmentation techniques to improve model robustness.
Significance: Automated tumor shape classification enables rapid and scalable analysis of morphological differences, providing valuable insights for cancer diagnostics and treatment planning.
