Acute gastrointestinal bleeding is an acute medical condition which warrants rapid diagnosis and management. It is one of the leading causes of high morbidity and mortality among patients presenting to the emergency department.¹ Methods of evaluation of suspected gastrointestinal bleed include endoscopy/colonoscopy, radionuclide imaging, CT angiography, CT enterography, catheter angiography, push endoscopy and capsule endoscopy etc.² None of the above methods at the present provide unequivocal information on quantity of GI bleed. The quantification and rate of bleed helps the clinician to assess the imminent clinical status and strategy for management. It helps to stratify patients into those with high or low rates of bleed, thus stratifying patients for urgent management.

CT angiography is an extremely useful and cost-effective modality in the assessment of gastrointestinal bleeding.³ Plain CT demonstrates bleed as areas of hyper-density in the lumen. With the use of intravenous contrast agents, the site of bleed can be localized and nature of the bleeder (arterial or venous) can be delineated. This information provides essential data for identification and access to the vessel causing the acute bleeding, thus planning the need for therapeutic embolization or bowel packing.

CT particularly has major advantages over other techniques in patients with severe bleeding.⁴ In some cases endoscopy or colonoscopy may be ineffective as the endoscope may not be passed up to the bleeding site for various reasons. Contexts include situations where bowel preparation is inadequate or in acute cases of bleeding wherein visibility of the bleeding vessel is compromised due to excessive blood products and clots in the bowel lumen.⁵ CT also serves as a versatile, fast initial modality as the investigation is completed quickly and large active bleeding can be identified easily along with additional anatomical information of the adjacent organs.

Three-phase examination consisting of unenhanced, arterial and portal venous phases is performed in CT angiography of abdomen in patients with acute gastrointestinal bleeding. The scan should include the complete abdomen and pelvis, from the diaphragm to the inferior pubic ramus. No oral contrast material is generally administered. Intravenous contrast material is administered through an antecubital vein with a power injector at a rate of 4 ml/sec, followed by a 50 ml of saline chaser. The dose is adjusted for body weight and iodine concentration, with the total volume of contrast material typically varying between 100 and 125 ml of an agent high in iodine concentration (>300 mg of iodine per ml).⁶

Our examinations were performed on a 128 slice scanner using helical acquisition of images. A plain CT was performed without oral contrast, followed by dual phase contrast examination with intravenous contrast (1.2-1.5 ml/kg) given at the rate of 4-5 ml/sec. Arterial phase images were obtained using automated bolus tracking with triggering ROI on upper abdominal aorta and delayed phase obtained 45-50 seconds later. Images were obtained at 120 Kv and 300-330 mAs for adults (100-120 kV and 120-150 mAs for the child). Slices were acquired at 5 mm thickness and reconstructed to 0.675 mm sections. Coronal and Sagittal reformatting were done. Visual inspection of plain and contrast images were done and using Philips Intellispace software version 11.0 (Koninklije Philips NV, Netherlands). The contrast extravasated area with a HU of more than 90 was seeded manually and its volume was estimated by the software. Based on the time elapsed between the phases, and estimated volume of contrast that has extravasated, the likely rate of bleeding was calculated.

We performed 3D volumetric assessment in three cases of massive bleeding. Follow on endoscopic examination showed significant blood clots, however no bleeder was identified. Quantitative estimation of clot volume was done by endoscopist, subjectively.

On work station volumetric assessment of extravasated contrast was done. DICOM images were opened using “CT viewer” application in Philips Intellispace version 11.0. After visual inspection of plain and contrast images, the contrast extravasated area with an HU of more than +90 was seeded manually using “Clip and 3D segmentation”. Its volume was calculated by the software. Using “Tissue management”, images were then volume rendered to produce a 3D display and volume was calculated. Generally volume estimation of bleed is made on a formula ‘ABC/2x slice thickness’ wherein A, B and C represent dimensions of selected attenuation profile. This formula is extensively used in the volume assessment of intracranial bleed.⁷ Rate of bleed is a guesstimate, done and documented with the help two data sets, i.e. arterial and portal venous phase. Arterial values are determined by the point of bolus triggering. Portal phase obtained after 50 seconds. Volume change elapsed between the two phases of CT acquisition when standardised for unit time, allows calculation of rate of bleeding. Estimated volumes were compared. Higher rate of bleed identified patients who are in need to urgent management.

MCECT is a very effective initial modality in the diagnosis of active gastrointestinal bleeding particularly in cases where endoscopy has failed to localize the site of bleed. The volumetric assessment of bleeding enhances the utility of CT, thus help in stratifying patients of differing severity of bleed. Such estimation of bleeding rate and volume has implications for the method of treatment and more importantly, the rapidity and urgency with which treatment is implemented.

Volumetric assessment of bleed or hematoma on abdominal imaging has not been explored adequately. CT- based method to quantify bleeding volume on plain CT has been explored in the study by Chikamatsu et al.⁸ in the context of post-biopsy renal bleeding. However we have not come across, on reviewing the literature, similar estimation of gastrointestinal bleed on MCECT. Detection of bleed on CT examination is based on the attenuation differences between bleeding focus and adjacent tissues. Haemoglobin has a higher Hounsfield Units (HU), hence can be detected if sufficient quantity is accumulated in organs where there is possibility for high intrinsic tissue contrast. Thus detection rate is highly variable depending upon the site, size, rate of bleed and suitable anatomy. Non-contrast CT estimation of bleed in GI bleed is ineffective due to small quantity and complex environment. Intravenous contrast can improve detection, (due to high HU of iodine) if detectable quantity is extravasated at bleeding site. Temporal changes in the quantity of detected bleed will help in estimating rate of bleeding. Improved reconstruction methods, computational techniques offered in the state of the art work stations facilitate predictable detection in a clinical setting. We have used Philips Intellispace WS, Ver 11 with facility for seeding an area of interest and region growing as defined by the set HU. Thus in our study we were able to predict the volume and rate of bleeding by setting the HU at 90+ unit. We have noted that rate of bleed can be estimated in low rates [3.8 cc/ min (0.063 cc/sec)] as well as high rates [40 cc/min (0.67 cc/sec)]. Temporal imaging with multiple phases is of value here to detect minimal amount of pooled contrast as an indicator of active haemorrhage. CT can visualize both intra-luminal and extra-luminal bleed. Estimated volume and rate of bleed provides significant additional insight in complex clinical scenarios. Estimated rate of bleed, along with clinical parameters provide good idea of the ongoing hemodynamic status.

The fact that no bowel preparation is needed for CT evaluation gives it a significant advantage over other modalities in evaluation of patients in acute state. The examination time required for CT is around 5 minutes with image-processing time of around 10-15 minutes. Thus the total examination time is usually around 15-20 minutes.

Clinical scoring systems for estimation of severity, risk of re-bleeding and mortality of upper gastrointestinal bleeding include Forrest classification, Rockall score, Baylor Bleeding score, Spanish Almela score and PNED score. These scoring methods comprise of components from patient history, clinical variables and endoscopic findings, with significant emphasis on endoscopic findings.⁹ Clinical risk scores are also used to identify high-risk patients requiring immediate intervention and low-risk patients who can be safely discharged. The major clinical scoring systems reported in literature are the clinical Rockall score and the Glasgow-Blatchford Score (GBS). Studies from several countries have suggested that the relatively simple GBS is superior to the Rockall score, at admission, in predicting clinically relevant end points.¹⁰

In clear contrast with upper GI bleeding, very few reports have focused on the prediction of outcomes in patients with lower GI bleeding. The existing scores like Oakland score, GBS, Rockall post endoscopy score, BLEED score, Strate score and NOBLADS score predict mortality, severity of bleeding and need for intervention. However data from these scoring systems are scarce and more studies are needed to establish its validity.¹¹

Our study has several limitations. CT detection methods are not very sensitive in detection of small bleed in anatomically complex regions. Low rates of bleeding may be overlooked and intermittent bleeding may not be detected by this method. This method also requires familiarity with volumetric assessment, though the learning curve is not steep. There are individual variations in rate of bleed and detection depending upon, Hg level, circulating blood volume, body surface area and general haemodynamic status of the patient during the examination.⁸ The CT estimation of rate of bleeding can be subjectively validated by endoscopy, surgery or angiography respectively as illustrated in our cases. However objective assessment of the same is not possible using the above methods. Since there is a need for urgent management, absolute estimation of bleeding rates using nuclear scintigraphy are also not useful.

Some discrepancy on the observations is expected in the context of bleed as it is progressive with a variable course. Time difference between CT and definitive therapy will alter clinical status. Our calculation is based on the assumption that rate of bleeding is constant, which need not be the case. Also our technique is more sensitive to bleed in the arterial phase, thus may underestimate slow venous (portal or systemic venous) bleeding. Another shortcoming of this technique is that metallic artifacts can potentially interfere with visualization of contrast in the bowel lumen on CT angiography and may lead to inaccurate assessment. Such limitations are well documented in CT angiography¹² Further limitations, though less relevant in acute setting include dose, and risks affiliated with contrast material such as allergy, nephropathy, or hyperthyreosis.³ There is a scope for material imaging in quantification of GI bleed.⁴ It allows more accurate of measurement of macromolecules and can image iodine as a separate entity when measuring rate of bleeding during CT angiography. Material imaging is likely to enhance the sensitivity of quantitative volume measurement by more accurately measuring subtle, slow bleed in the venous phases.

The estimation of bleed volume on CT proves superior to clinical scoring systems as it can identify and stratify the rate of bleeding regardless of its aetiology, site and whether bleeding is intra or extra luminal.

Volumetric estimation of GI bleed on MCECT is a novel concept in the evaluation of gastrointestinal bleeding, with an immense potential. Methods identify the bleed site with prediction on rate of bleed, thus help stratifying patient for management. We hope that with routine use of material imaging and more user friendly software options for volume estimation, this technique will be a standard practice in emergency GI bleed management.