When all else fails - Radiological management of severe gastrointestinal bleeding

Artificial Intelligence for Identification of Images with Active Bleeding in Mesenteric and Celiac Arteries Angiography

PURPOSE

The purpose of this study is to evaluate the efficacy of an artificial intelligence (AI) model designed to identify active bleeding in digital subtraction angiography images for upper gastrointestinal bleeding.

METHODS

Angiographic images were retrospectively collected from mesenteric and celiac artery embolization procedures performed between 2018 and 2022. This dataset included images showing both active bleeding and non-bleeding phases from the same patients. The images were labeled as normal versus images that contain active bleeding. A convolutional neural network was trained and validated to automatically classify the images. Algorithm performance was tested in terms of area under the curve, accuracy, sensitivity, specificity, F1 score, positive and negative predictive value.

RESULTS

The dataset included 587 pre-labeled images from 142 patients. Of these, 302 were labeled as normal angiogram and 285 as containing active bleeding. The model's performance on the validation cohort was area under the curve 85.0 ± 10.9% (standard deviation) and average classification accuracy 77.43 ± 4.9%. For Youden's index cutoff, sensitivity and specificity were 85.4 ± 9.4% and 81.2 ± 8.6%, respectively.

CONCLUSION

In this study, we explored the application of AI in mesenteric and celiac artery angiography for detecting active bleeding. The results of this study show the potential of an AI-based algorithm to accurately classify images with active bleeding. Further studies using a larger dataset are needed to improve accuracy and allow segmentation of the bleeding.

Efficacy of transcatheter arterial embolization in treating nonvariceal gastric remnant bleeding: a retrospective 5-year study

BACKGROUND

Gastric remnant bleeding is a special case of upper gastrointestinal bleeding with certain specific disease characteristics, and some matters of transcatheter arterial embolization (TAE) for hemostasis need attention. In this study, we aimed to explore the clinical use of TAE in patients with nonvariceal gastric remnant bleeding and identify the factors influencing the clinical efficacy of these interventions.

METHODS

Data were retrospectively analyzed from 42 patients for whom angiography and embolization were performed but could not be treated endoscopically or had failed endoscopic management in our department between January 2018 and January 2023 due to nonvariceal gastric remnant bleeding. We investigated the relationship between the incidence of re-bleeding and the following variables: sex, age, pre-embolization gastroscopy/contrast-enhanced computer tomography, embolization method, aortography performance, use of endoscopic titanium clips, and the presence of collateral gastric-supplying arteries.

RESULTS

Forty-two patients underwent 47 interventional embolizations. Of these, 16 were positive for angiographic findings, and 26 were negative. Based on arteriography results, different embolic agents were selected, and the technical success rate was 100%. The incidence of postoperative re-bleeding was 19.1% (9/47), and the overall clinical success rate was 81.0% (34/42). Logistic regression analysis of the relationship between the incidence of early re-bleeding following embolization and the proportion of collateral gastric supply arteries revealed an odds ratio of 10.000 (p = 0.014).

CONCLUSIONS

Utilizing TAE for nonvariceal gastric remnant bleeding is safe and effective. The omission of collateral gastric-supplying arteries can lead to early re-bleeding following an intervention.

Post-Embolization Hemoglobin Changes: When to Consider Re-intervention

PURPOSE

Decline in hemoglobin (Hgb) levels is common post-embolization, but there is no consensus on the classification of patients based on risk for re-bleeding or re-intervention. The current study evaluated post-embolization Hgb level trends with the goal of understanding the factors predictive of re-bleeding and re-intervention.

MATERIALS AND METHODS

All patients who underwent embolization for gastrointestinal (GI), genitourinary, peripheral, or thoracic arterial hemorrhage from 01/2017 to 01/2022 were reviewed. Data included demographics, periprocedural pRBC transfusion (TF) or pressor requirements, and outcome. Lab data consisted of Hgb values pre-embolization, immediately post-embolization, and daily values on days 1-10 after embolization. Hgb trends were compared between patients across TF and re-bleeding outcomes. Regression model was used to examine factors predictive of re-bleeding and magnitude of Hgb reduction post-embolization.

RESULTS

A total of 199 patients were embolized for active arterial hemorrhage. Perioperative Hgb level trends were similar for all sites and between TF + and TF- patients, showing a decline reaching a nadir within 6 days post-embolization followed by an upward trend. Maximum Hgb drift was predicted by GI embolization (p = 0.018), TF before embolization (p = 0.001), and use of vasopressor (p = 0.000). Patients with Hgb drop  > 15% within the first two days post-embolization had a higher chance of having a re-bleeding episode (p = 0.04).

CONCLUSION

Perioperative Hgb trends showed a consistent downward drift followed by an upward shift, irrespective of TF requirement status or site of embolization. Using a cut-off value of 15% Hgb reduction within the first two days post-embolization may be helpful to assess re-bleeding risk.

ENDOVASCULAR DIAGNOSTICS AND TREATMENT OF HEMORRHAGES IN MILITARY AND CIVILIAN PATIENTS FOLLOWING ABDOMINAL SURGERY

OBJECTIVE

The aim: The retrospective analysis of the angiographic picture and clinical results of endovascular treatment of patients with postoperative abdominal and gastrointestinal bleeding was carried out.

PATIENTS AND METHODS

Materials and methods: The results of 447 endovascular diagnostic and therapeutic interventions in 391 patients (46 military and 345 civilians) with postoperative bleeding performed from 2012 to 2022 were studied. According to computer tomography with contrast enhancement, the source of bleeding was identified in 216 (67.7%) cases.

RESULTS

Results: In 345 (88.2%) patients, it was possible to reliably identify the source of bleeding on angiography. In 46 (11.8%) patients with an unexplained source of bleeding, the target arterial pool was determined on the basis of localization, volume, and features of surgical intervention and considered as preventive interventions. A total of 447 endovascular hemostatic interventions were performed on 391 patients. A stent graft was installed in 27 patients, 420 embolization were performed in 364 patients. Thus, in 43 (11.0%) patients, embolization was performed repeatedly, in 12 cases - three times, in 1 case - four times. In 16 cases (15 cases of prophylactic embolization), endovascular hemostasis was ineffective and required subsequent surgical intervention.

CONCLUSION

Conclusions: Endovascular interventions are an effective method of diagnosis and treatment of postoperative abdominal bleeding. Prophylactic embolization allows you to prevent the recurrence of postoperative bleeding with an instrumentally undiagnosed source, however, you need to be prepared for the multi-stage treatment aimed at sequentially shutting down the collateral blood supply to the damaged area.

Endovascular Treatment of Gastrointestinal Hemorrhage

Background and Objectives: Severe non-variceal gastrointestinal bleeding is a life-threatening condition with complicated treatment if endoscopic therapy fails. In such cases, transcatheter arterial embolization is recommended. The technical and clinical effects of this technique were analyzed in this group of patients, as well as its complication rate and 30-day mortality. Materials and Methods: Patient data over a one-decade period (from 2010 to 2019) were analyzed retrospectively; 27 patients (18 men and 9 women; median age 61 years) treated by endovascular embolization in our institution, with clinically significant gastrointestinal hemorrhage after unsuccessful or impossible endoscopic treatment, were identified, and their data were collected. Results: The source of bleeding was found in 88% of patients, but embolization was performed in 96% of them. The overall technical success rate was 96.8%, and the clinical success was 88.5%. Re-bleeding occurred in eight cases, five of whom had re-embolization that was technically successful in four cases. The incidence of re-bleeding was significantly higher in patients with two or more comorbidities (p = 0.043). There was one serious complication (4%) in the group, and minor difficulties occurred in 18% of patients; 30-day mortality reached 22%. Mortality was significantly higher in the group of patients with re-bleeding (p = 0.044). Conclusions: Transcatheter arterial embolization is a mini-invasive method with high technical success in patients with endoscopically untreatable gastrointestinal bleeding; it is also suitable for high-risk cases. Mortality (to a significant extent) depends on the occurrence of re-bleeding and the patient’s comorbidities.

Current Status and Future Perspective of Artificial Intelligence in the Management of Peptic Ulcer Bleeding: A Review of Recent Literature

With the decreasing incidence of peptic ulcer bleeding (PUB) over the past two decades, the clinician experience of managing patients with PUB has also declined, especially for young endoscopists. A patient with PUB management requires collaborative care involving the emergency department, gastroenterologist, radiologist, and surgeon, from initial assessment to hospital discharge. The application of artificial intelligence (AI) methods has remarkably improved people's lives. In particular, AI systems have shown great potential in many areas of gastroenterology to increase human performance. Colonoscopy polyp detection or diagnosis by an AI system was recently introduced for commercial use to improve endoscopist performance. Although PUB is a longstanding health problem, these newly introduced AI technologies may soon impact endoscopists' clinical practice by improving the quality of care for these patients. To update the current status of AI application in PUB, we reviewed recent relevant literature and provided future perspectives that are required to integrate such AI tools into real-world practice.

Relationship between angiography timing and angiographic visualization of extravasation in patients with acute non-variceal gastrointestinal bleeding

Background

Angiographic embolization is now considered the first-line therapy for acute gastrointestinal (GI) bleeding refractory to endoscopic therapy. The success of angiographic embolization depends on the detection of the bleeding site. This study aimed to identify the clinical and procedural predictors for the angiographic visualization of extravasation, including angiography timing, as well as analyze the outcomes of angiographic embolization according to the angiographic visualization of extravasation.

Methods

The clinical and procedural data of 138 consecutive patients (mean age, 66.5 years; 65.9% men) who underwent angiography with or without embolization for acute non-variceal GI bleeding between February 2008 and July 2018 were retrospectively analyzed.

Results

Of the 138 patients, 58 (42%) had active extravasation on initial angiography and 113 (81.9%) underwent embolization. The angiographic visualization of extravasation was significantly higher in patients with diabetes (p = 0.036), a low platelet count (p = 0.048), high maximum heart rate (p = 0.002) and AIMS65 score (p = 0.026), upper GI bleeding (p = 0.025), and short time-to-angiography (p = 0.031). The angiographic embolization was successful in all angiograms, with angiographic visualization of extravasation (100%). The clinical success of patients without angiographic visualization of extravasation (83.9%) was significantly higher than that of patients with angiographic visualization of extravasation (65.5%) (p = 0.004). In multivariate analysis, the time-to-angiography (odds ratio 0.373 [95% CI 0.154–0.903], p = 0.029) was the only significant predictor associated with the angiographic visualization of extravasation. The cutoff value of time-to-angiography was 5.0 h, with a sensitivity and specificity of 79.3% and 47.5%, respectively (p = 0.012).

Conclusions

Angiography timing is an important factor that is associated with the angiographic visualization of extravasation in patients with acute GI bleeding. Angiography should be performed early in the course of bleeding in critically ill patients.