Detection of Pancreatic Cancer in CT Scan Images Using PSO SVM and Image Processing

Area of Science:

• Medical Imaging and Diagnostics
• Computational Biology and Machine Learning
• Oncology

Background:

• Pancreatic cancer has a very low five-year survival rate, making early detection crucial.
• Medical imaging, such as CT scans, aids in identifying abnormalities but faces accessibility challenges due to high costs.
• Advanced image processing and machine learning techniques are needed for efficient and accessible cancer detection.

Purpose of the Study:

• To present a novel method for detecting pancreatic cancer in CT scan images.
• To evaluate the effectiveness of machine learning algorithms, specifically Particle Swarm Optimization Support Vector Machine (PSO SVM), in conjunction with image processing techniques for this task.
• To compare the performance of PSO SVM against other classification algorithms like Naive Bayes and AdaBoost.

Main Methods:

• Image preprocessing using the Gaussian elimination filter for noise removal.
• Image segmentation employing the K-means algorithm for partitioning and identifying regions of interest.
• Feature extraction using Principal Component Analysis (PCA) and classification using PSO SVM, Naive Bayes, and AdaBoost.

Main Results:

• The Gaussian elimination filter effectively removed noise from CT scan images.
• Image segmentation and PCA facilitated the extraction of relevant features for classification.
• The PSO SVM algorithm achieved superior accuracy, sensitivity, and specificity in detecting pancreatic cancer compared to Naive Bayes and AdaBoost.

Conclusions:

• The proposed method combining image processing and PSO SVM shows significant promise for accurate and potentially more accessible early detection of pancreatic cancer.
• PSO SVM is a highly effective classification algorithm for analyzing medical images in oncology.
• Further development and validation could lead to improved diagnostic tools for pancreatic cancer.