Scholarlink

<p>In order to improve the accuracy of cancer diagnosis, a&#xa0;new convolutional neural network architecture is presented, which provides automatic segmentation and detection of kidney tumors on three-dimensional images obtained by computed tomography. The proposed approach is based on the integration of three complementary technologies: multilevel convolutional processing, residual connections, and U‑Net architectural principles. This approach ensures efficient processing of 3D medical data. An original neural network system for the segmentation of kidney images obtained by computed tomography and detection of kidney tumors has been created. To validate this system, a&#xa0;dataset was used that satisfied the following criteria: sufficient sample size (at least 300 cases); expert image labeling; verified diagnoses; diverse clinical cases; and open access for research purposes. The KiTS19 dataset (Kidney Tumor Segmentation 2019), provided by the University of Minnesota Clinic through the Grand Challenge platform, provides the most complete satisfaction of the above criteria. The dataset includes 300 labeled 3D images of kidneys obtained by computed tomography with confirmed diagnoses. The experimental stages include dataset preprocessing, including normalization and augmentation; system training based on 210 cases; and validation by using an independent sample consisting of 90&#xa0;cases. The experimental results demonstrate high diagnostic efficiency of the developed system, such as the accuracy of automatic segmentation of anatomical kidney structures (96% based on the Dice coefficient), and the accuracy of tumor detection and segmentation (91% based on the Dice coefficient). The obtained results can be used in the following areas of clinical practice: preoperative planning and navigation during organ-preserving surgeries; automated screening of CT images for early detection of kidney tumors; quantitative assessment of the dynamics of tumor growth when monitoring the course of the disease; and support of clinical decision-making in oncourology.</p>

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

Artificial intelligence in oncourology: integrated deep learning technologies in the tasks of segmentation of three-dimensional images of kidney tumors

  • 期刊论文

  • 发表日期：

    2025年2月1日

查看全文
  • V. G. Nikitaev,
  • D. Yu. Pushkar,
  • V. B. Matveev,
  • A. N. Pronichev,
  • O. V. Nagornov,
  • A. I. Otchenashenko,
  • A. I. Kleyman

摘要

In order to improve the accuracy of cancer diagnosis, a new convolutional neural network architecture is presented, which provides automatic segmentation and detection of kidney tumors on three-dimensional images obtained by computed tomography. The proposed approach is based on the integration of three complementary technologies: multilevel convolutional processing, residual connections, and U‑Net architectural principles. This approach ensures efficient processing of 3D medical data. An original neural network system for the segmentation of kidney images obtained by computed tomography and detection of kidney tumors has been created. To validate this system, a dataset was used that satisfied the following criteria: sufficient sample size (at least 300 cases); expert image labeling; verified diagnoses; diverse clinical cases; and open access for research purposes. The KiTS19 dataset (Kidney Tumor Segmentation 2019), provided by the University of Minnesota Clinic through the Grand Challenge platform, provides the most complete satisfaction of the above criteria. The dataset includes 300 labeled 3D images of kidneys obtained by computed tomography with confirmed diagnoses. The experimental stages include dataset preprocessing, including normalization and augmentation; system training based on 210 cases; and validation by using an independent sample consisting of 90 cases. The experimental results demonstrate high diagnostic efficiency of the developed system, such as the accuracy of automatic segmentation of anatomical kidney structures (96% based on the Dice coefficient), and the accuracy of tumor detection and segmentation (91% based on the Dice coefficient). The obtained results can be used in the following areas of clinical practice: preoperative planning and navigation during organ-preserving surgeries; automated screening of CT images for early detection of kidney tumors; quantitative assessment of the dynamics of tumor growth when monitoring the course of the disease; and support of clinical decision-making in oncourology.