Improved fusion imaging for the detection of Endometriosis

A method to improve fusion (MRI/US) image quality for endometriosis detection

Key Points

Importance:

• Fusion imaging, a combined magnetic resonance imaging (MRI) and ultrasonography (US), may play an important role in the detection of endometriosis.

Highlights:

• A novel algorithm is suggested to improve the detection of endometriosis, by the fusion of two different radiological techniques.

What’s done here:

• This article introduces a new fusion method, magnetic resonance, and ultrasound imaging, for better detection and accurate diagnosis of endometriosis.

Key Results:

• Fusion imaging using both magnetic resonance imaging and ultrasound raises new challenges that require various algorithms to improve signal-to-noise in the resulting image and ensure the best possible image quality.
• Two formation models were used to account for the drawbacks of each imaging modality such as the low spatial resolution on the US, and non-linear cost function when fusing both modalities.

Limitations: 

• The techniques elucidated in this study need to be applied in imaging cases of endometriosis and compared to standard imaging protocols.

Lay Summary

This paper studied a new fusion method for magnetic resonance (MR) and ultrasound (US) imaging for the appropriate diagnosis of endometriosis.

The diagnosis of endometriosis in patients with a chronic pelvic disease is delayed, due to the disease, patient, and physician-led factors. A recent active study area is how ordering physicians can work together to diagnosis endometriosis earliest possible. However possible screening and imaging methods are still not the best by the fact that the heterogeneous nature of the endometriosis.

Although laparoscopy continues to be the gold standard for the diagnosis and treatment of endometriosis, endometriosis surgeons are increasingly relying on MRI studies before the surgery in order to accurately and efficiently localize the disease distribution to make an appropriate surgical plan. 

Image fusion is to gather important information from multiple images and gather the information into one single image which is generally more informative than previous pictures. However, this requires solving several technical challenges. Endometriosis is a typical example of a disease pathology that requires the use of MR and US modalities in conventional clinical practice.

This study, performed by El Mansouri et al. from the University of Toulouse, France, presents a new fusion imaging technique using and combining the advantages of good contrast and signal to noise ratio of magnetic resonance imaging and the good spatial resolution for the ultrasonography. To solve the inverse problems of performing a super-resolution of the magnetic resonance and denoising of the ultrasonography, a non-linear polynomial function is used by a proximal alternating linearized minimization algorithm. Finally, a simulation for the detection of endometriosis using a phantom study was performed, and the accuracy of the algorithm was qualitatively and quantitatively shown.

This work is a first attempt for fusing magnetic resonance and ultrasonography images and opens several interesting perspectives. Further investigation, the progression of the technique, and learning the functional dependence between images are needed to apply the method elucidated in this study for fusion detection of endometriosis and its better applicability to in vivo data.

This paper is recently published in the journal named "IEEE Transactions on Image Processing" 

Research Source: https://pubmed.ncbi.nlm.nih.gov/32142435