AI Case Study
Qure.ai achieves 90% accuracy using deep learning to diagnose pulmonary consolidation in chest x-rays
Qure.ai has received medical device certification for its qXR product which analyses X-rays using neural networks to detect tuberculosis. Clinicians are able to upload images and receive automatic reports to aid their diagnoses. According to initial testing, their method is 90% accurate.
Industry
Healthcare
Healthcare Providers And Services
Project Overview
Qure.ai has "developed qXR, a chest x-ray product that can identify 15 of the most common chest x-ray abnormalities. On Thursday, qXR received CE certification from the European Medical Device Directives, clearing the way for commercialization in 32 European countries" according to VentureBeat. "Clinics have two choices when it comes to installation: a cloud-hosted setup in which radiologists digitize and upload scans to Qure.ai’s servers for analysis, or a locally hosted, on-premise solution that uses off-the-shelf hardware". Diagnostic reports of x-ray abnormalities are automatically then generated for clinicians to use.
From Qure.ai's blog, "At Qure, we build deep learning models to detect abnormalities from radiological images. These models require huge amount of labeled data to learn to diagnose abnormalities from the scans. So, we collected a large dataset from several centers, which included both in-hospital and outpatient radiology centers. These datasets contain scans and the associated clinical radiology reports. For now, we use radiologist reports as the gold standard as we train deep learning algorithms to recognize abnormalities on radiology images. While this is not ideal for many reasons... it is currently the most scalable way to supply classification algorithms with the millions of images that they need in order to achieve high accuracy."
Reported Results
According to VentureBeat, a preliminary study demonstrated that "qXR was more than 90 percent accurate in correctly diagnosing tuberculosis". From Qure.ai's validation section of its website, it states that the method described in ""Efficacy of Deep Learning for Screening Pulmonary Tuberculosis" achieved an AUC-ROC of 0.88, which is considered good.
Technology
From the abstract for "Efficacy of Deep Learning for Screening Pulmonary Tuberculosis": "x-rays with consolidation are identified from their reports using natural language processing techniques. Images are preprocessed to a standard size and normalised to remove source dependency. These images are trained using deep residual neural networks. Multiple models are trained on various selective subsets of the dataset along with one model trained on entire data set. Scores yielded by each of these models is passed through a 2-layer neural network to generate final probabilities for presence of consolidation in an x-ray."
Classification for the images taken from radiologist reports is done using rules-based natural language processing, and thus is technically not considered AI.
Function
Operations
General Operations
Background
According to VentureBeat over "10.4 million people were infected with tuberculosis in 2016, according to the Center for Disease Control and Prevention. Of those, 1.7 million died from resulting complications, many in developing regions of the world with limited access to radiology departments." The Qure.ai website states that "[c]hest x-rays are widely used to identify pulmonary consolidation because they are highly accessible, cheap and sensitive. Automating the diagnosis in chest x-rays can reduce diagnostic delay, especially in resource-limited settings."
Benefits
Data
In Qure.ai's paper "Efficacy of Deep Learning for Screening Pulmonary Tuberculosis" a "dataset of 423,218 chest x-rays with corresponding reports (collected from 166 centres across India spanning 22 x-ray machine variants from 9 manufacturers) is used for training and validation."