Keep breathing! Common motion helps multi-modal mapping.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Accurate information transfer between multi-modal medical images is challenging, especially when images are acquired from different regions or at different times.
• Existing methods often require precise spatial overlap, limiting their applicability in free-breathing acquisitions.

Purpose of the Study:

• To develop and evaluate an unconventional approach for transferring information between multi-modal images by exploiting common temporal signals.
• To enable accurate image feature localization without requiring the same anatomical region to be captured by all modalities.

Main Methods:

• The proposed method extracts a low-dimensional description of image sequences from different modalities (e.g., MRI, US).
• It identifies a common cause signal, such as breathing, to synchronize the sequences.
• Similar sub-sequences are identified to predict the spatial location of image features.

Main Results:

• The approach was tested on 2D MRI and 2D ultrasound (US) liver image sequences from three volunteers.
• Slow feature analysis was used for signal extraction, achieving an average prediction error of 2.6 mm for same-time acquisitions and 4.2 mm for different-time acquisitions.
• Quantitative evaluation demonstrated the method's ability to predict MRI feature locations accurately.

Conclusions:

• The developed method offers an effective way to transfer information between multi-modal images acquired during free-breathing, even without overlapping regions.
• Exploiting the common breathing signal provides a robust synchronization mechanism for image analysis.
• This approach has potential applications in improving image-guided interventions and diagnostics.