Volumetric dynamic MR-defecography
Valentin Ratz1, Tobias Wech1, Andreas Schindele2, Alexander Sauer1, Alexander Dierks1, Joachim Reibetanz3, Alfio Borzi2, Thorsten Bley1, and Herbert Köstler1

1Institute of Diagnostic and Interventional Radiology, Unisversity Hospital Würzburg, Würzburg, Germany, 2Institute of Mathematics, University Würzburg, Würzburg, Germany, 3Department of General-, Visceral-, Vascular- and Pediatric Surgery, Unisversity Hospital Würzburg, Würzburg, Germany

Synopsis

Current 2D MR-defecography techniques lack information about lateral localized pathologies. The purpose of this project was to implement a dynamic 3D MR-defecography which covers a larger area and thereby gathers more detailed information on the anatomy during defecation. Therefore we implemented a radial 3D TrueFISP Stack of Stars sequence with density-weighting and view-sharing. The highly undersampled datasets, which we have acquired in 5 patients, were reconstructed using a modified FISTA compressed sensing algorithm. The proposed method allows for a dynamic 3D examination of the fast defecation process. Therefore the detection of lateral localized pathologies is possible as well.

Purpose

Epidemiological studies have found that the estimated incidence of chronic constipation is up to 27% of the general population [1]. The gold standard to evaluate affected patients is the dynamic Entero-Colpo-Cysto-Defecography [2]. In clinical routine also 2D MR-defecography is performed, but only one to three 2D slices at a temporal footprint of about one second are acquired [2,3]. This leads to limitations in the detection of lateral localized pathologies. Therefore it was our aim to implement a dynamic 3D MR-defecography which covers a larger area and thereby gathers more detailed information on the anatomy during defecation.

Methods

A radial 3D TrueFISP “Stack-of-Stars” [4] sequence was implemented (see Fig. 1 left). Density-weighting was implemented in z-direction to sample the center of the k-space more densely than the periphery (see Fig. 1 middle). Additionally, a view-sharing scheme was implemented to almost double the number of reconstructed time frames compared to the standard sampling scheme (see Fig. 1 right). The sequence was applied to the first 5 patients in an ongoing study. Each time frame featured a 3D block with seven partitions at a spatial resolution of 1.3x1.3x4.0mm3 to 2.0x2.0x6.0mm3 and a temporal footprint of 1.3s to 2.0s. Undersampled data (acceleration factor with respect to a fully sampled Cartesian sampling ≈ 6) were reconstructed by Compressed Sensing [5] applying a modified FISTA algorithm [6]. 3D dynamic MR-defecography was performed on a whole body 3T MR system (Siemens, MAGNETOM Prisma). The rectum of the patients was opacified with 200ml of sonographic gel. During the measurement the patient was asked to strain and squeeze several times and then to evacuate the rectum.

Results

After the compressed sensing reconstruction of all time frames, the defecation process can be visualized in all partitions of the 3D block. Figure 2 shows eight time frames from the time series of partition four of the dynamic 3D MR-defecography. It is clearly visualized how an anterior rectocele (red arrow) evolves during defecation. Figure 3 shows the seven acquired partitions of time frame four. Hence the lateral extent of the rectocele can be visualized as well.

Discussion

The presented images show that the proposed method allows for a dynamic 3D examination of the fast defecation process. The anatomical structure of the pelvic floor during defecation can be visualized and analyzed. Each 3D volume has a temporal footprint < 2s and the volume is updated due to the view-sharing with a rate of about 1s.

Conclusion

Current 2D MR-defecography lacks information about lateral localized pathologies. Volumetric dynamic imaging of the defecation process allows for the visualization and interpretation of pelvic floor disorders and additionally allows for the detection of lateral localized pathologies.

Acknowledgements

No acknowledgement found.

References

[1] Andromanakos et al. Eur J Gastroenterol Hepatol 2015; 27(3) 204-214

[2] Cappabianca et al. Int J Colorectal Dis 2011; 26:1191–1196

[3] Seynaeve et al. JBR–BTR 2006; 89:182-189

[4] Block et al. J Korean Soc Magn Reson Med 2014; 18(2): 87-106

[5] Lustig et al. Magn Reson Med 2007; 58(6):1182-95

[6] Beck A et al. SIIMS 2009; 2(1):183-202

Figures

Standard Stack of Stars sampling scheme (left) and the two applied modifications density-weighting (middle) and view-sharing (right).

Temporal evolution of a rectocele (red arrow) during defecation.

All seven partitions of timeframe four in Figure 2. The lateral extent of the anterior rectocele (red arrow) can be visualized.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2967