Noise analysis in real-time portal imaging. I. Quantization noise

Area of Science:

• Medical Physics
• Radiological Imaging
• Image Processing

Background:

• Video-based real-time portal imaging is crucial for radiation therapy guidance.
• Quantization noise can limit image quality in digital imaging systems.
• Understanding noise sources is essential for optimizing image acquisition protocols.

Purpose of the Study:

• To investigate the dominant source of quantization noise in video-based real-time portal imaging.
• To compare the effects of digital frame averaging versus digital frame addition on image quality.
• To determine the impact of bit depth on signal-to-noise ratio and contrast resolution.

Main Methods:

• Analysis of quantization noise sources in integer frame averaging and analog-to-digital conversion (A/D).
• Acquisition and processing of portal images using a Newvicon camera, with varying frame averaging and addition techniques.
• Evaluation of signal-to-noise ratio (SNR) and contrast resolution with different bit depths (8-bit vs. 16-bit).

Main Results:

• Quantization noise from truncation during frame averaging dominates over A/D noise, especially when averaging over 64 frames.
• Digital frame addition mitigates truncation noise, allowing for higher achievable signal-to-noise ratios compared to frame averaging.
• An 8-bit A/D with digital frame averaging limits SNR to 886.8, while 16-bit processing with frame addition improves contrast resolution.

Conclusions:

• Digital frame addition is superior to frame averaging for minimizing quantization noise in video-based real-time portal imaging.
• Utilizing 16-bit processing with digital frame addition significantly enhances image quality, offering better contrast resolution than 8-bit processing with frame averaging.
• Optimizing frame processing techniques and bit depth is critical for maximizing image quality in portal imaging systems.