Real time MRI prostate segmentation based on wavelet multiscale products flow tracking

Area of Science:

• Medical Imaging
• Oncology
• Signal Processing

Background:

• Prostate cancer is a leading cause of male cancer deaths in North America.
• Early detection and treatment significantly improve patient survival rates.
• Combined Magnetic Resonance Imaging and Spectroscopic Imaging (MRI/MRSI) are vital for early prostate cancer detection, but rely heavily on accurate region of interest (ROI) determination.

Purpose of the Study:

• To present a novel method for segmenting the prostate in MRI datasets.
• To enhance the performance of prostate MRI/MRSI techniques through improved segmentation.
• To facilitate earlier and more accurate prostate cancer detection.

Main Methods:

• The proposed method utilizes the distinct behavior of signal singularities and noise in the wavelet domain to detect prostate borders.
• Prostate contour tracing is achieved using spatially variant rules informed by prior knowledge of the prostate's general shape.
• The technique focuses on segmenting the entire prostate for improved ROI definition in MRI/MRSI.

Main Results:

• The novel segmentation method demonstrated promising accuracy in identifying prostate borders.
• The approach effectively distinguishes between signal singularities and noise in the wavelet domain.
• Successful application to clinical datasets indicates the method's practical utility.

Conclusions:

• Accurate prostate segmentation is essential for optimizing MRI/MRSI performance in early prostate cancer detection.
• The presented wavelet-domain method offers a robust approach to prostate segmentation.
• This technique has the potential to significantly enhance the diagnostic capabilities of prostate imaging for improved patient outcomes.