Artificial Intelligence Enhances Diagnostic Accuracy of Contrast Enemas in Hirschsprung Disease Compared to Clinical Experts

Abstract

Introduction

Contrast enema (CE) is widely used in the evaluation of suspected Hirschsprung disease (HD). Deep learning is a promising tool to standardize image assessment and support clinical decision-making. This study assesses the diagnostic performance of a deep neural network (DNN), with and without clinical data, and compares its interpretation with that of pediatric surgeons and radiologists.

Materials and Methods

In this retrospective study, 1,471 CE images from patients <15 years were analyzed, with 218 images used for testing. A DNN, pediatric radiologists, and surgeons independently reviewed the testing set, with and without clinical data. Diagnostic performance was assessed using ROC and PR curves, and interobserver agreement was evaluated using Fleiss' kappa. Rectal biopsy served as the reference standard.

Results

The DNN achieved high diagnostic accuracy (area under the receiver operating characteristic curve [AUC-ROC] = 0.87) in CE interpretation, with improved performance when combining anteroposterior and lateral images (AUC-ROC = 0.92). Clinical data integration further enhanced model sensitivity and negative predictive value. The super-surgeon (majority voting of colorectal surgeons) outperformed most individual clinicians (sensitivity 81.8%, specificity 79.1%), while the super-radiologist (majority voting of radiologists) showed moderate accuracy. Interobserver analysis revealed strong agreement between the model and surgeons (Cohen's kappa = 0.73), and overall consistency among experts and the model (Fleiss' kappa = 0.62).

Conclusions

Artificial intelligence-assisted CE interpretation achieved higher specificity and comparable sensitivity to that of the clinicians. Its consistent performance and substantial agreement with experts support its potential role in improving CE assessment in HD.

Keywords

Hirschsprung disease - contrast enema - artificial intelligence - deep learning - interobserver agreement

• References

• 1 Das K, Mohanty S. Hirschsprung disease - current diagnosis and management. Indian J Pediatr 2017; 84 (08) 618-623
• 2 Kyrklund K, Sloots CEJ, de Blaauw I. et al. ERNICA guidelines for the management of rectosigmoid Hirschsprung's disease. Orphanet J Rare Dis 2020; 15 (01) 164
• 3 Zani A, Eaton S, Morini F. et al.; EUPSA Network Office. European Paediatric Surgeons' Association survey on the management of Hirschsprung disease. Eur J Pediatr Surg 2017; 27 (01) 96-101
• 4 Diamond IR, Casadiego G, Traubici J, Langer JC, Wales PW. The contrast enema for Hirschsprung disease: Predictors of a false-positive result. J Pediatr Surg 2007; 42 (05) 792-795
• 5 Putnam LR, John SD, Greenfield SA. et al. The utility of the contrast enema in neonates with suspected Hirschsprung disease. J Pediatr Surg 2015; 50 (06) 963-966
• 6 Peyvasteh M, Askarpour S, Ostadian N, Moghimi MR, Javaherizadeh H. Diagnostic accuracy of barium enema findings in Hirschsprung's disease. Arq Bras Cir Dig 2016; 29 (03) 155-158
• 7 Frongia G, Günther P, Schenk JP. et al. Contrast enema for Hirschsprung disease investigation: Diagnostic accuracy and validity for subsequent diagnostic and surgical planning. Eur J Pediatr Surg 2016; 26 (02) 207-214
• 8 Shen D, Wu G, Suk H-I. Deep learning in medical image analysis. Annu Rev Biomed Eng 2017; 19: 221-248
• 9 Goldstein E, Keidar D, Yaron D. et al. COVID-19 Classification of X-Ray Images Using Deep Neural Networks Key Results. Eur Radiol 2021; 31 (12) 9654-9663
• 10 Zhou Z, Qian X, Hu J. et al. An artificial intelligence-assisted diagnosis modeling software (AIMS) platform based on medical images and machine learning: a development and validation study. Quant Imaging Med Surg 2023; 13 (11) 7504-7522
• 11 Huang SG, Qian XS, Cheng Y, Guo WL, Zhou ZY, Dai YK. Machine learning-based quantitative analysis of barium enema and clinical features for early diagnosis of short-segment Hirschsprung disease in neonate. J Pediatr Surg 2021; 56 (10) 1711-1717
• 12 Alehossein M, Roohi A, Pourgholami M, Mollaeian M, Salamati P. Diagnostic accuracy of radiologic scoring system for evaluation of suspicious Hirschsprung disease in children. Iran J Radiol 2015; 12 (02) e12451
• 13 Madabhushi A, Lee G. Image analysis and machine learning in digital pathology: Challenges and opportunities. Med Image Anal 2016; 33: 170-175
• 14 Ashraf R, Habib MA, Akram M. et al. Deep convolution neural network for big data medical image classification. IEEE Access 2020; 8: 105659-105670
• 15 Duci M, Magoni A, Santoro L. et al. Enhancing diagnosis of Hirschsprung's disease using deep learning from histological sections of post pull-through specimens: preliminary results. Pediatr Surg Int 2023; 40 (01) 12
• 16 Hwang TJ, Servaes S, Mattei P, Anupindi SA. Radiologist performance in the interpretation of contrast enemas performed for Hirschsprung's disease in children >1 year of age. Clin Radiol 2017; 72 (06) 519.e11-519.e19