Proton magnetic resonance spectroscopy as a potential tool for differentiating between abdominal fluid collections

To determine the utility of proton magnetic resonance spectroscopy (MRS) in distinguishing abdominal fluid types.

Abdominal fluid samples were obtained from patients undergoing therapeutic percutaneous drainage. In vitro spectroscopy was performed using a 1.5-T scanner and a head coil. Single voxel spectra were obtained using a point resolved spin-echo sequence with water suppression (TR/TE 2000 msec/35 msec). The peak pattern for each sample was examined and the signal-to-noise ratio (SNR) estimated (ratio of tallest peak to noise at <0 ppm).

Thirty-five samples were analyzed: purulent collection (eight), serosanguinous collection (eight), non-chylous ascites (six), chylous ascites (one), bile (seven), and bile with iodinated contrast media (five). The mean SNR of the dominant peak was: purulent collection, 12.7; serosanguinous collection, 3.2; non-chylous ascites, 2.4; chylous ascites, 8.8; bile, 1.4; and bile with contrast media, 60.8. Pus samples had a broad based peak pattern with continuous signal of >1.5 ppm width situated within the range 0.2-2.5 ppm, not found in other samples. Chylous ascites (one sample) had a distinctive peak at 1.2 ppm. Bile with contrast had three peaks at 3.5/3.6, 2.6, and 2.1 ppm. No other patterns were found to be discriminatory. Common non-specific patterns seen included a bifid peak at 1.1-1.3 ppm and a broad based peak situated between 3 and 4 ppm.

The H1 spectra of purulent fluid has a higher SNR than common non-purulent abdominal fluids and a distinct broad based peak pattern from 0.2-2.5 ppm. Proton spectroscopy may be a useful tool for distinguishing purulent from non-purulent intra-abdominal collections.