【阿里matlab算法】matlab实现基于K-means、模糊C-means和优化K-means的乳腺肿瘤分割研究——肿瘤分割

    % 读取图像
    I = imread('breast_tumor.png');
    I_gray = rgb2gray(I); % 灰度化
    
    % 预处理（滤波和阈值化）
    I_filtered = medfilt2(I_gray, [3 3]); % 中值滤波
    I_binary = imbinarize(I_filtered, 'adaptive', 'ForegroundPolarity', 'dark', 'Sensitivity', 0.4); % 自适应阈值化
    
    % K-means聚类分割
    k = 3; % 设定聚类数量
    [idx, C] = kmeans(double(I_filtered(:)), k);
    I_kmeans = reshape(idx, size(I_filtered));
    I_tumor_kmeans = I_kmeans == 1; % 假设肿瘤区域为聚类1
    
    % 模糊C-means聚类分割
    m = 2; % 模糊因子
    k = 3; % 设定聚类数量
    [centers, U, objFcn] = fcm(double(I_filtered(:)), k, [2, 100, 1e-5, 0], m);
    I_fcm = reshape(max(U, [], 2), size(I_filtered));
    I_tumor_fcm = I_fcm == 1; % 假设肿瘤区域为聚类1
    
    % DBSCAN聚类分割
    epsilon = 15; % 邻域半径
    minPts = 5; % 邻居数量
    I_vector = double(I_filtered(:));
    labels = DBSCAN(I_vector, epsilon, minPts);
    I_dbscan = reshape(labels, size(I_filtered));
    I_tumor_dbscan = I_dbscan == 1; % 假设肿瘤区域为聚类1
    
    % 显示结果
    figure;
    subplot(2, 2, 1);
    imshow(I_gray);
    title('灰度图像');
    
    subplot(2, 2, 2);
    imshow(I_tumor_kmeans);
    title('K-means分割');
    
    subplot(2, 2, 3);
    imshow(I_tumor_fcm);
    title('模糊C-means分割');
    
    subplot(2, 2, 4);
    imshow(I_tumor_dbscan);
    title('DBSCAN分割');
    
    % DBSCAN函数定义（需要额外添加）
    function labels = DBSCAN(X, epsilon, minPts)
    % DBSCAN聚类算法实现
    % X: 数据集（向量形式）
    % epsilon: 邻域半径
    % minPts: 邻居数量
    % labels: 聚类标签
    
    % 初始化变量
    n = size(X, 1);
    labels = zeros(n, 1);
    cluster_id = 0;
    
    for i = 1:n
    if labels(i) == 0
    % 找到邻居
    neighbors = regionQuery(X, i, epsilon);
    if numel(neighbors) < minPts
    % 噪声点
    labels(i) = -1;
    else
    cluster_id = cluster_id + 1;
    labels = expandCluster(X, labels, i, neighbors, cluster_id, epsilon, minPts);
    end
    end
    end
    end
    
    function labels = expandCluster(X, labels, i, neighbors, cluster_id, epsilon, minPts)
    % 扩展聚类
    labels(i) = cluster_id;
    k = 1;
    while k <= numel(neighbors)
    point = neighbors(k);
    if labels(point) == -1
    labels(point) = cluster_id;
    elseif labels(point) == 0
    labels(point) = cluster_id;
    new_neighbors = regionQuery(X, point, epsilon);
    if numel(new_neighbors) >= minPts
    neighbors = [neighbors; new_neighbors]; %#ok<AGROW>
    end
    end
    k = k + 1;
    end
    end
    
    function neighbors = regionQuery(X, i, epsilon)
    % 区域查询
    dists = sqrt(sum((X - X(i, :)).^2, 2));
    neighbors = find(dists <= epsilon);
    end
    
    
    c
    
    
![](https://ad.itadn.com/c/weblog/blog-img/images/2025-08-17/CqpPoYmHL16bnXVS3FKTWDEvr9RZ.png)