Scholarlink

Analysis of clinical notes is a popular as well as challenging field in Natural language processing, especially in the healthcare domain. The handwritten text dataset is an image dataset that further needs to be processed. Hence, primarily we need to extract the textual part from the images properly. Identifying the specific disease condition from clinical notes thus can be elaborated as a junction of image processing and natural language processing. Prediction of cancer disease conditions from clinical notes can prevent doctors from overlooking possible risks. Doctors can advise patients about taking hospital admission. Also, the patient can plan for further treatment accordingly. According to domain experts, after providing the treatment, the cancer condition can be classified into four different categories. Those are complete response, progressive disease, partial response, and stable disease. In our proposed approach, we need to start with an Optical Character Recognition (OCR) engine to extract proper text information from handwritten notes. Further, the NLP process was executed. Finally, a deep neural network is used to classify different cancer disease conditions. Our approach is compared to some recent implementations, and we achieved more promising results compared to them.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Investigation of Handwritten Cancer Notes Classification Using Transformer-Based Deep Neural Network Pipeline

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tiyasa Chakraborty,
  • Ashok Kumar Bhadra,
  • Debashis Nandi

摘要

Analysis of clinical notes is a popular as well as challenging field in Natural language processing, especially in the healthcare domain. The handwritten text dataset is an image dataset that further needs to be processed. Hence, primarily we need to extract the textual part from the images properly. Identifying the specific disease condition from clinical notes thus can be elaborated as a junction of image processing and natural language processing. Prediction of cancer disease conditions from clinical notes can prevent doctors from overlooking possible risks. Doctors can advise patients about taking hospital admission. Also, the patient can plan for further treatment accordingly. According to domain experts, after providing the treatment, the cancer condition can be classified into four different categories. Those are complete response, progressive disease, partial response, and stable disease. In our proposed approach, we need to start with an Optical Character Recognition (OCR) engine to extract proper text information from handwritten notes. Further, the NLP process was executed. Finally, a deep neural network is used to classify different cancer disease conditions. Our approach is compared to some recent implementations, and we achieved more promising results compared to them.