Clinical Scorecard: A Multiphasic CT-Based Decision Support Model for Pre-Treatment Differentiation of Pediatric Hepatoblastoma and Focal Nodular Hyperplasia

At a Glance

Key Highlights

• DL+QEF model achieved an AUC of 0.980 and 94.2% accuracy in differentiating HB from FNH.

• Junior radiologist accuracy improved from 88.4% to 95.0% with model assistance.

• The model integrates high sensitivity for HB with high specificity for FNH.

• DL+QEF outperformed single-phase DL models in diagnostic accuracy.

• Potential utility in atypical imaging presentations.

Guideline-Based Recommendations

Diagnosis

• Utilize multiphasic CT for initial evaluation of suspected HB and FNH.

• Incorporate serum AFP levels in diagnostic evaluation.

Management

• Surgical excision for confirmed HB; conservative management for FNH unless symptomatic.

Monitoring & Follow-up

• Regular imaging follow-up for patients with FNH to assess for changes.

Risks

• Risk of misdiagnosis leading to unnecessary interventions in FNH.

Patient & Prescribing Data

Children undergoing evaluation for liver lesions.

Neoadjuvant chemotherapy may be indicated for HB to enhance resectability.

Clinical Best Practices

• Employ deep learning models to assist in the interpretation of multiphasic CT.

• Consider atypical imaging presentations when diagnosing pediatric liver lesions.

Related Resources & Content

• American Academy of Pediatrics: Guidelines for the Diagnosis and Management of Pediatric Liver Tumors

• Journal of Pediatric Surgery: Recent Advances in the Management of Hepatoblastoma

• Radiology Society of North America: Imaging Techniques for Pediatric Liver Lesions

• Children's Oncology Group: Treatment Protocols for Hepatoblastoma