Optimized Tree Strategy with Principal Component Analysis Using Feature Selection-Based Classification for Newborn...

Area of Science:

• Medical Imaging
• Computational Pathology
• Neonatology

Background:

• Newborn jaundice grading is a critical yet challenging area in medical research.
• Accurate mitotic count is essential for determining jaundice severity in newborns.
• Traditional methods for jaundice detection can be time-consuming and less precise.

Purpose of the Study:

• To develop an automated system for mitotic detection and jaundice grading in newborns using histopathology images.
• To improve the accuracy and efficiency of neonatal hyperbilirubinemia assessment.
• To investigate the impact of feature selection on classification performance.

Main Methods:

• Utilized principal component analysis (PCA) for feature selection.
• Employed an optimal tree strategy classifier for automatic mitotic detection.
• Applied image processing techniques to analyze real-time and benchmark histopathology datasets.

Main Results:

• The developed system demonstrated accurate, quick, and time-efficient results for newborn jaundice detection.
• Image processing techniques were found to be critical for predicting neonatal hyperbilirubinemia.
• The combination of feature selection and classification models addressed performance bottlenecks.

Conclusions:

• Automated mitotic detection and grading using PCA and optimal tree classifiers offer a promising approach for neonatal jaundice assessment.
• Image processing is vital for accurate prediction and diagnosis of neonatal hyperbilirubinemia.
• The proposed method provides a reliable and efficient alternative to conventional jaundice grading techniques.