Improved early stroke detection: wavelet-based perception enhancement of computerized tomography exams

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Nonenhanced computerized tomography (CT) is used for acute stroke detection by identifying hypodense areas.
• Subtle signs of infarction can be challenging to perceive on standard CT scans, potentially leading to delayed diagnosis.

Purpose of the Study:

• To propose and evaluate a novel wavelet-based image processing method for improving the detection of acute stroke on nonenhanced CT.
• To enhance the perception of subtle hypodense areas indicative of infarction, thereby increasing diagnostic efficiency.

Main Methods:

• A wavelet-based image processing technique was developed for data denoising and local contrast enhancement in a multi-scale domain.
• Initial segmentation of brain tissue and extraction of regions susceptible to density changes were employed for more effective data processing.
• The method was experimentally verified on patient data.

Main Results:

• The proposed method significantly enhanced the visibility of subtle hypodensities, often invisible in standard CT reviews.
• Sensitivity of stroke diagnosis increased substantially to 56.3% compared to 12.5% with standard CT scan preview.
• The image processing approach demonstrated improved detection efficiency for perceptual ischemic changes.

Conclusions:

• Wavelet-based image processing offers a promising approach to enhance subtle signs of acute stroke on nonenhanced CT scans.
• This method can significantly improve the sensitivity and efficiency of stroke diagnosis, aiding in earlier detection and treatment.
• Further validation and integration into clinical workflows could enhance acute stroke management.