AML detection from bone marrow microscopic images using CNN

The detection of Acute Myeloid Leukemia (AML) from bone marrow microscopic images is a crucial step in early diagnosis and treatment planning. This project presents an efficient AML detection system utilizing image processing and machine learning techniques. The proposed system processes bone marrow microscopic images by resizing them to 256x256 pixels, converting them to grayscale, and performing noise removal using a median filter. Chi-square feature extraction is employed to capture statistical characteristics by dividing the image into 8x8 blocks and computing the chi-square values for each block's intensity distribution. The extracted features are then classified using two models: K-Nearest Neighbors (KNN) and Logistic Regression. The KNN classifier achieved promising accuracy, demonstrating its robustness in distinguishing between AML-positive and AML-negative samples. The Logistic Regression model further improved classification performance by accurately predicting cancerous and non-cancerous samples through binary classification. The system effectively integrates feature extraction with machine learning models, ensuring high accuracy and reliability for AML detection. This approach holds significant potential for aiding medical professionals in diagnosing AML efficiently and improving patient outcomes. The implemented system offers a scalable solution for automated leukemia detection using microscopic image data.

Keywords: AML Disease Dataset, Image Processing Techniques, Machine Learning Techniques, K-Nearest Neighbors (KNN) and Logistic Regression, and Accuracy.