Deep Learning Magnetic Resonance Imaging Radiomics for Diagnostic Value of Hepatic Tumors in Infants

Hepatic tumors in the perinatal period are associated with significant morbidity and mortality in affected patients. The conventional diagnostic tool, such as alpha-fetoprotein (AFP) shows limited value in diagnosis of infantile hepatic tumors. This retrospective-prospective study is aimed to evaluate the diagnostic efficiency of the deep learning system through analysis of magnetic resonance imaging (MRI) images before initial treatment.

Study Overview

• Status: Recruiting
• Conditions: Hepatoblastoma; Hepatic Hemangioendothelioma
• Intervention / Treatment: Diagnostic test: Radiomic Algorithm

Detailed Description

Hepatic tumors seldom occur in the perinatal period. They comprise approximately 5% of the total neoplasms of various types occurring in the fetus and neonate. Infantile hemangioendothelioma is the leading primary hepatic tumor followed by hepatoblastoma. It should be mentioned that alpha-fetoprotein (AFP) is highly elevated during the first several months after birth even in normal infants, thus the diagnostic value of AFP is limited for infantile patients with hepatic tumors. This study is a retrospective-prospective design by West China Hospital, Sichuan University, including clinical data and radiological images. A retrospective database was enrolled for patients with definite histological diagnosis and available magnetic resonance imaging (MRI) images from June 2010 and December 2020. The investigators have constructed a deep learning radiomics diagnostic model on this retrospective cohort and validated it internally. A prospective cohort would recruit infantile patients diagnosed as liver tumor since January 2021. The proposed deep learning model would also be validated in this prospective cohort externally. The established model would be able to assist diagnosis for hepatic tumor in infants.

• Study Type: Observational
• Enrollment (Anticipated): 200

Contacts and Locations

Study Locations

• China
  • Sichuan
    • Chendu, Sichuan, China, 610041
      • Recruiting
      • West China Hospital, Sichuan University

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 1 second to 1 year (Child)
• Accepts Healthy Volunteers: No
• Genders Eligible for Study: All

• Sampling Method: Non-Probability Sample

Study Population

Patients who had liver tumor and completed the abdominal MRI examination before operation, biopsy, neoadjuvant chemotherapy, and radiotherapy.

Description

Inclusion Criteria:

• Age between newborn and 12 months
• Receiving no treatment before diagnosis
• With written informed consent

Exclusion Criteria:

• Clinical data missing
• Unavailable MRI images
• Without written informed consent

Study Plan

How is the study designed?

Design Details

• Observational Models: Cohort
• Time Perspectives: Other

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
Retrospective cohort; The internal cohort was retrospectively enrolled in West China Hospital, Sichuan University from June 2010 and December 2020. It is a training and internal validation cohort.	Diagnostic test: Radiomic Algorithm; Different radiomic, machine learning, and deep learning strategies for radiomic features extraction, sorting features and model constriction.
Prospective cohort; The same inclusion/exclusion criteria were applied for the same center prospectively. It is an external validation cohort.	Diagnostic test: Radiomic Algorithm; Different radiomic, machine learning, and deep learning strategies for radiomic features extraction, sorting features and model constriction.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
The diagnostic accuracy of infantile liver tumors with deep learning algorithm	The diagnostic accuracy of infantile liver tumors with deep learning algorithm.	1 month

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
The diagnostic sensitivity of infantile liver tumors with deep learning algorithm	The diagnostic sensitivity of infantile liver tumors with deep learning algorithm.	1 month
The diagnostic specificity of infantile liver tumors with deep learning algorithm	The diagnostic specificity of infantile liver tumors with deep learning algorithm.	1 month
The diagnostic positive predictive value of infantile liver tumors with deep learning algorithm	The diagnostic positive predictive value of infantile liver tumors with deep learning algorithm.	1 month
The diagnostic negative predictive value of infantile liver tumors with deep learning algorithm	The diagnostic negative predictive value of infantile liver tumors with deep learning algorithm.	1 month

Collaborators and Investigators

• Sponsor: West China Hospital

Study record dates

Study Major Dates

• Study Start (Actual): January 1, 2021
• Primary Completion (Anticipated): December 31, 2023
• Study Completion (Anticipated): December 31, 2023

Study Registration Dates

• First Submitted: December 8, 2021
• First Submitted That Met QC Criteria: December 22, 2021
• First Posted (Actual): December 27, 2021

Study Record Updates

• Last Update Posted (Actual): December 27, 2021
• Last Update Submitted That Met QC Criteria: December 22, 2021
• Last Verified: December 1, 2021

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Digestive System Diseases; Neoplasms by Histologic Type; Neoplasms; Neoplasms by Site; Digestive System Neoplasms; Liver Diseases; Neoplasms, Complex and Mixed; Hemangioma; Neoplasms, Vascular Tissue; Liver Neoplasms; Hepatoblastoma; Hemangioendothelioma
• Other Study ID Numbers: HX2021-345

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: No

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No