Detection and segmentation of intracranial hematoma in CT image(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2021年第9期
- Page:
- 1090-1096
- Research Field:
- 医学人工智能
- Publishing date:
Info
- Title:
- Detection and segmentation of intracranial hematoma in CT image
- Author(s):
- LIU Yuanfeng1; LIU Libo1; LIU Yun2
- 1. School of Information Engineering, Ningxia University, Yinchuan 750021, China 2. Department of Radiology, Cardio- and Cerebrovascular
Disease Hospital, General Hospital of Ningxia Medical University, Yinchuan 750002, China
- Keywords:
- edge detection genetic algorithm region-growing algorithm hematoma detection and segmentation
- PACS:
- R318；TP391
- DOI:
- 10.3969/j.issn.1005-202X.2021.09.008
- Abstract:
- Objective To propose an intracranial hematoma-like noise detection method based on improved Canny operator for
improving the accuracy of hematoma segmentation. Methods The brain tissues were firstly segmented by region-growing
algorithm, and the interference information such as skull was removed. Then the hematoma-like noise at the edge of brain
was detected by improved Canny edge detection, and the noise was eliminated by an and operation with the original image.
Finally, the genetic algorithm based on OTSU fitness function was used to segment intracranial hematoma accurately.
Results In 200 randomly selected brain hematoma images, the accuracy of hematoma detection was 96.3%, and Dice
similarity reached 93.5%. Conclusion The proposed method can be used to detect and segment intracranial hematoma
accurately and effectively.
Last Update: 2021-09-27