Performance evaluation of coherence-based adaptive imaging using clinical breast data

Diagnostic ultrasound often suffers from reduced image quality due to variations in sound speed within human tissue. These inhomogeneities distort signals, which negatively impacts both spatial resolution and overall contrast. Prior research has shown that adaptive imaging techniques can mitigate these distortions by adjusting signal processing parameters. However, no prior work had resolved the performance of coherence-based weighting in complex clinical breast environments. Previous studies primarily relied on phantom models, leaving a gap regarding real-world diagnostic efficacy. That uncertainty drove the need for testing these algorithms on actual patient data. Conventional correlation-based methods frequently struggle with model assumptions and hardware limitations during routine examinations. This study addresses those challenges by applying a coherence-based approach to clinical breast imaging datasets.

The aim of this study was to evaluate the performance of a coherence-based adaptive imaging approach using clinical breast data. Researchers sought to address the persistent issue of sound-velocity inhomogeneities in diagnostic ultrasound. These variations typically degrade both spatial resolution and image contrast during routine clinical procedures. The team specifically compared their proposed method against a standard correlation-based technique documented in existing literature. This comparison was motivated by the known limitations of correlation-based models, including their reliance on near-field phase-screen assumptions. Furthermore, the lack of two-dimensional arrays often introduces significant integration errors in traditional systems. By testing the coherence-based method on patient data, the authors intended to validate its effectiveness in a real-world clinical setting. This work provides a necessary assessment of how adaptive weighting can improve diagnostic clarity for various breast conditions.

The review approach involved testing a coherence-based algorithm on clinical breast datasets acquired from twenty-five patients. Investigators utilized a programmable ultrasound system featuring a 5 MHz, 128-channel linear transducer array. This design allowed for the direct comparison of the proposed method against a widely recognized correlation-based technique. The team processed raw channel data to calculate pixel weights based on signal coherence. They avoided the restrictive near-field, phase-screen models that often hinder traditional correlation-based processing. By applying this adaptive weighting, the researchers assessed improvements in contrast ratio and contrast-to-noise ratio metrics. The analysis included various lesion types to ensure a comprehensive evaluation of the imaging performance. Finally, the team measured changes in object dimensions for specific cases to quantify spatial resolution gains.

Key findings from the literature demonstrate that the coherence-based method achieves an average contrast ratio improvement of 8.57 dB. This result significantly exceeds the 0.42 dB gain observed with the correlation-based approach. Additionally, the coherence-based technique yielded a 23.2% improvement in the contrast-to-noise ratio. In contrast, the correlation-based method provided only a 3.35% increase in this metric. For a milk-of-calcium case, the coherence-based approach improved contrast by 4.47 dB. Spatial resolution also improved, with axial dimensions shrinking from 0.39 mm to 0.32 mm. Lateral dimensions similarly decreased from 0.51 mm to 0.43 mm using the proposed adaptive weighting. These metrics confirm the efficacy of the coherence-based algorithm across diverse clinical breast pathologies.

The authors demonstrate that coherence-based weighting significantly enhances image quality in clinical breast examinations. This approach provides superior contrast improvements compared to traditional correlation-based strategies. The findings suggest that this technique effectively handles signal distortions without requiring complex phase-screen models. By utilizing standard linear arrays, the method remains practical for existing diagnostic ultrasound systems. The data indicates substantial gains in both contrast ratio and contrast-to-noise ratio metrics. These results confirm the utility of the proposed algorithm for characterizing various breast lesions. The researchers propose that this adaptive weighting strategy offers a robust alternative for improving diagnostic accuracy. Future clinical implementation could benefit from the increased clarity provided by this coherence-driven processing framework.