改成物件導向
該程式的主要任務是從指定的文字檔案中提取 ROI（感興趣區域）資訊，統計不同 ROI 標籤（如 56，63，69 ……）的出現次數，並繪製統計結果的條形圖。透過將功能模組化到 RoiAnalyzer 類中，程式碼變得更加結構化和可維護。
【文字檔案】：
- 原始碼
import re import numpy as np import seaborn as sns import matplotlib.pyplot as plt def ret_roi_value_dict(txt_path): output = [] line_number = 0 with open(txt_path, 'r') as file: for line in file: line_number += 1 # 使用正規表示式提取case_name和pixel_value match = re.match(r'Case\s+(.*?)\s+has\s+\[(.*?)\]', line) if match: case_name = match.group(1) # 去除字串中的多餘空格，並進行分割 pixel_value = list(map(int, match.group(2).strip().split())) output.append({'case_name': case_name, 'pixel_value': pixel_value}) else: print(f"Line {line_number} not matched: {line.strip()}") return output def show_roi_value(roi_value_dict, fig_save_path, save_dpi): # 提取所有 "pixel_value" 對應的值 all_pixel_values = [] for item in roi_value_dict: all_pixel_values.extend(item["pixel_value"]) # 計算每個元素的出現次數 unique_elements, counts = np.unique(np.array(all_pixel_values), return_counts=True) # 將 unique_elements 和 counts 轉換為 list，並根據 counts 進行排序 sorted_indices = np.argsort(counts)[::-1] sorted_unique_elements = unique_elements[sorted_indices] sorted_counts = counts[sorted_indices] # 定義前五個顏色 top_colors = ['red', 'blue', 'green', 'yellow', 'purple'] bar_colors = ['grey'] * len(unique_elements) # 將排名第二到第六的元素的顏色改為指定顏色 for i in range(1, min(6, len(sorted_unique_elements))): # 找到 sorted_unique_elements[i] 在 original unique_elements 中的位置 original_index = np.where(unique_elements == sorted_unique_elements[i])[0][0] bar_colors[original_index] = top_colors[i-1] # 繪製條形圖 plt.figure(figsize=(10, 6)) bars = sns.barplot(x=unique_elements, y=counts, palette=bar_colors) plt.title('TB ROI Information') plt.xlabel('ROI label') plt.ylabel('Case count') plt.xticks(rotation=45) # 旋轉 x 軸標籤以提高可讀性 plt.tight_layout() # 確保標籤不會被裁剪 plt.grid(axis="y") # 在每個 bar 上方標註數量 for i, bar in enumerate(bars.patches): height = bar.get_height() plt.text(bar.get_x() + bar.get_width() / 2, int(height), f'{int(height)}', ha='center', va='bottom', color='black') # 儲存影像 plt.savefig(fig_save_path, dpi=save_dpi) def main(): keyword_list = ["MaXin_50_Newlabel_final", "MaXin_100", "ZhangRui_50_Newlabel_final", "ZhangRui_54"] roi_value_dict = [] for keyword in keyword_list: temp = ret_roi_value_dict(txt_path=f"/homes/xchang/Projects/Multi-center_Tuberculosis/segmentation/nnUNet_dl/{keyword}.txt") roi_value_dict.extend(temp) show_roi_value(roi_value_dict=roi_value_dict, fig_save_path=f"/homes/xchang/Projects/Multi-center_Tuberculosis/roi_new_values_final.png", save_dpi=600) if __name__ == "__main__": main() - 最佳化後
import re import numpy as np import seaborn as sns import matplotlib.pyplot as plt class RoiAnalyzer: def __init__(self, keywords, base_path): self.keywords = keywords self.base_path = base_path self.roi_value_dict = [] def ret_roi_value_dict(self, txt_path): output = [] line_number = 0 with open(txt_path, 'r') as file: for line in file: line_number += 1 # 使用正規表示式提取case_name和pixel_value match = re.match(r'Case\s+(.*?)\s+has\s+\[(.*?)\]', line) if match: case_name = match.group(1) # 去除字串中的多餘空格，並進行分割 pixel_value = list(map(int, match.group(2).strip().split())) output.append({'case_name': case_name, 'pixel_value': pixel_value}) else: print(f"Line {line_number} not matched: {line.strip()}") return output def collect_data(self): for keyword in self.keywords: txt_path = f"{self.base_path}/{keyword}.txt" temp = self.ret_roi_value_dict(txt_path) self.roi_value_dict.extend(temp) def show_roi_value(self, fig_save_path, save_dpi): # 提取所有 "pixel_value" 對應的值 all_pixel_values = [] for item in self.roi_value_dict: all_pixel_values.extend(item["pixel_value"]) # 計算每個元素的出現次數 unique_elements, counts = np.unique(np.array(all_pixel_values), return_counts=True) # 將 unique_elements 和 counts 轉換為 list，並根據 counts 進行排序 sorted_indices = np.argsort(counts)[::-1] sorted_unique_elements = unique_elements[sorted_indices] sorted_counts = counts[sorted_indices] # 定義前五個顏色 top_colors = ['red', 'blue', 'green', 'yellow', 'purple'] bar_colors = ['grey'] * len(unique_elements) # 將排名第二到第六的元素的顏色改為指定顏色 for i in range(1, min(6, len(sorted_unique_elements))): # 找到 sorted_unique_elements[i] 在 original unique_elements 中的位置 original_index = np.where(unique_elements == sorted_unique_elements[i])[0][0] bar_colors[original_index] = top_colors[i-1] # 繪製條形圖 plt.figure(figsize=(10, 6)) bars = sns.barplot(x=unique_elements, y=counts, palette=bar_colors) plt.title('TB ROI Information') plt.xlabel('ROI label') plt.ylabel('Case count') plt.xticks(rotation=45) # 旋轉 x 軸標籤以提高可讀性 plt.tight_layout() # 確保標籤不會被裁剪 plt.grid(axis="y") # 在每個 bar 上方標註數量 for i, bar in enumerate(bars.patches): height = bar.get_height() plt.text(bar.get_x() + bar.get_width() / 2, int(height), f'{int(height)}', ha='center', va='bottom', color='black') # 儲存影像 plt.savefig(fig_save_path, dpi=save_dpi) def run(self, fig_save_path, save_dpi=600): self.collect_data() self.show_roi_value(fig_save_path, save_dpi) def main(): keywords = ["MaXin_50_Newlabel_final", "MaXin_100", "ZhangRui_50_Newlabel_final", "ZhangRui_54"] base_path = "/homes/xchang/Projects/Multi-center_Tuberculosis/segmentation/nnUNet_dl" fig_save_path = "/homes/xchang/Projects/Multi-center_Tuberculosis/roi_new_values_final.png" analyzer = RoiAnalyzer(keywords, base_path) analyzer.run(fig_save_path) if __name__ == "__main__": main()
