A Novel Cancer Diagnosis Framework Using Optimal Point Region Growing Segmentation and Pseudo-Zernike Moments
CSEF · 2015 Mammalian Biology Third Award
Overview
Objectives/Goals Cancer is the second most common cause of death in the US. Early diagnosis is the key to cure cancer and increase the patient's survival rate. The current computer aided diagnosis (CADx) system usually needs several iterations for segmenting the lesion and requires the extraction of many features for classification. It also lacks a common framework. The objective of this project is to design and develop a CADx common framework to simplify the diagnosis process and more efficiently and accurately diagnose cancers in their early stages. Methods/Materials The framework consists of automatic segmentation of a lesion using Optimal Point Region Growing Segmentation, reconstruction of the segmented lesion using Pseudo-Zernike moments, and Supported Vector Machine (SVM) classification of the lesion using the single feature, Root Mean Square (RMS) of Pseudo-Zernike moments. This novel CADx framework was implemented using MATLAB and validated on the mammographic images and the dermoscopic images from the Mammographic Image Analysis Society (MIAS) database and the published study examples, respectively. Results A comparative study among the various algorithms was performed on the selected mammographic images and dermoscopic images. The results demonstrated that the newly developed framework has over 86% average recognition rate, further improved the accuracy of Tahmasbi#s CADx algorithm by 7.56%, and reduced the False Negative Rate (FNR) and False Positive Rate (FPR) by 7.24% and 3.48%, respectively. Conclusions/Discussion This study has developed a novel CADx common framework to simplify the diagnosis process and more efficiently and accurately diagnose breast cancer and melanoma from a mammographic image and dermoscopic image, respectively. It has improved the best benchmark of recognition rate by 8%. The new framework is generic and can be used for diagnosing different types of cancers.
Summary statement
This study has not only developed a novel CADx framework to more efficiently and accurately diagnose breast cancer and melanoma but also laid a foundation for diagnosing other type of cancers.
Help received
Dr. James Li helped on the selection of the mammographic image database and provided feedback for my project. He also recommended "Digital Image Processing using MATLAB " by Rafael Gonzalez, which helped me to master my MATLAB image processing skills.
Awards (1)
Competition history
- CSEF 2015
Resources
Related projects
CSEF · 2014
A Novel Breast Cancer Detection Algorithm Using Point Region Growing Segmentation and Pseudo-Zernike Moments
ISEF · 2015
Improved Cancer Detection and Diagnosis through a Novel Combination of Cell Segmentation and Artificial Intelligence Techniques
ISEF · 2018
Machine Learning Approach to Computer Assisted Diagnosis of Skin Diseases
CSEF · 2014
Accurate Detection of Skin Cancer Using Multi Stage Neural Networks
CSEF · 2015
DermatoScan: Machine Vision, Analysis, Learning & Natural Computing Optimizations for the Early Detection of Skin Cancer
CSEF · 2014
Quantitative Analysis of Macrocellular Biomarkers in Early Stage DCIS Cytopathology Images, Using Machine Learning
ISEF · 2020
Improving Tumor Detection Accuracy in Mammograms of Dense Tissue Breast Cancer Patients with Machine Learning Methods and Image Processing Techniques
CSEF · 2018
A Deep Learning Approach for Analyzing Tumor Histopathological Image Slides to Predict Breast Cancer
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: California Science & Engineering Fair public projects