Potential of Deep Learning in Assessing Pneumoconiosis Depicted on Digital Chest Radiography

Completed

• Conditions: Pneumoconiosis
• Interventions: Other: convolutional neural networks (CNNs)

• Registration Number: NCT04963348
• Lead Sponsor: Peking University Third Hospital

Brief Summary

Pneumoconiosis is relatively prevalent in low/middle-income countries, and it remains a challenging task to accurately and reliably diagnose pneumoconiosis. The investigators implemented a deep learning solution and clarified the potential of deep learning in pneumoconiosis diagnosis by comparing its performance with two certified radiologists. The deep learning demonstrated a unique potential in classifying pneumoconiosis.

Detailed Description

The investigators retrospectively collected a dataset consisting of 1881 chest X-ray images in the form of digital radiography. These images were acquired in a screening setting on subjects who had a history of working in an environment that exposed them to harmful dust. Among these subjects, 923 were diagnosed with pneumoconiosis, and 958 were normal. To identify the subjects with pneumoconiosis, the investigators applied a classical deep convolutional neural network (CNN) called Inception-V3 to these image sets and validated the classification performance of the trained models using the area under the receiver operating characteristic curve (AUC).

Study Timeline

• Study Start: 2015-01-01
• Primary Completion: 2018-12-31
• Study Completion: 2019-12-31
• Status Verified: 2021-06
• Study First Submitted: 2021-06-28
• Study First Submitted QC: 2021-07-08
• Last Update Submitted: 2021-07-08
• Study First Posted: 2021-07-15
• Last Update Posted: 2021-07-15

Recruitment & Eligibility

• Status: COMPLETED
• Sex: All
• Target Recruitment: 1881

Inclusion Criteria

• industrial workers with a history of exposure to dust and underwent DR screening of pneumoconiosis from 2015 to 2018

Read More

Exclusion Criteria

• patients with poor image quality
• patients with incomplete clinical data

Read More

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Arm && Interventions

Group	Intervention	Description
convolutional neural network (CNN)	convolutional neural networks (CNNs)	a classical deep convolutional neural network (CNN) called Inception-V3 was applied to the image sets and validated the classification performance of the trained models

Primary Outcome Measures

Name	Time	Method
the diagnosis of pneumoconiosis	up to 6 months	The diagnosis and staging of pneumoconiosis were made by an expert panel consisting of certified radiologists and occupational physicians. The diagnosis of pneumoconiosis was confirmed by medical history and previous medical records(chest X-rays and pulmonary function testing).