Study Shows How Deep Learning and AI Diagnose TB
|
By MedImaging International staff writers Posted on 24 Apr 2017 |
|
Image: A chest X-Ray of a patient with active TB, and an X-Ray with a heat map overlay showing some of the results of the AI analysis (Photo courtesy of RSNA).
Researchers have found that they can use an artificial intelligence technique called deep learning to identify cases of tuberculosis on chest X-Ray exams with a net accuracy rate of 96%.
According to the World Health Organization (WHO) around 1.8 million people died from tuberculosis (TB) in 2016. A simple chest X-Ray exam can help radiologists identify the disease, but many TB patients live in remote areas without access to expert radiologists who can interpret the images, and diagnose the disease.
The study was carried out by researchers at the Thomas Jefferson University Hospital who trained artificial intelligence models to identify TB on chest X-rays. The goal of the research was to help screen and evaluate patients in TB-prevalent areas lacking access to radiologists. The study was published in the April 25, 2017, online issue of the journal Radiology.
The researchers used 1,007 X-Ray exams of patients with and without active TB for the study. The multiple TB-positive and TB-negative X-Ray datasets were used to train two different Deep Convolutional Neural Network (DCNN) models called AlexNet and GoogLeNet. The researchers found that the best performing Artificial Intelligence (AI) model was when both AlexNet and GoogLeNet were used together, resulting in a net accuracy of 96%.
Co-author of the study, Paras Lakhani, MD at TJUH, said, “There is a tremendous interest in artificial intelligence, both inside and outside the field of medicine. An artificial intelligence solution that could interpret radiographs for presence of TB in a cost-effective way could expand the reach of early identification and treatment in developing nations. The relatively high accuracy of the deep learning models is exciting. The applicability for TB is important because it’s a condition for which we have treatment options. It’s a problem that can be solved. We hope to prospectively apply this in a real world environment. An artificial intelligence solution using chest imaging can play a big role in tackling TB.”
According to the World Health Organization (WHO) around 1.8 million people died from tuberculosis (TB) in 2016. A simple chest X-Ray exam can help radiologists identify the disease, but many TB patients live in remote areas without access to expert radiologists who can interpret the images, and diagnose the disease.
The study was carried out by researchers at the Thomas Jefferson University Hospital who trained artificial intelligence models to identify TB on chest X-rays. The goal of the research was to help screen and evaluate patients in TB-prevalent areas lacking access to radiologists. The study was published in the April 25, 2017, online issue of the journal Radiology.
The researchers used 1,007 X-Ray exams of patients with and without active TB for the study. The multiple TB-positive and TB-negative X-Ray datasets were used to train two different Deep Convolutional Neural Network (DCNN) models called AlexNet and GoogLeNet. The researchers found that the best performing Artificial Intelligence (AI) model was when both AlexNet and GoogLeNet were used together, resulting in a net accuracy of 96%.
Co-author of the study, Paras Lakhani, MD at TJUH, said, “There is a tremendous interest in artificial intelligence, both inside and outside the field of medicine. An artificial intelligence solution that could interpret radiographs for presence of TB in a cost-effective way could expand the reach of early identification and treatment in developing nations. The relatively high accuracy of the deep learning models is exciting. The applicability for TB is important because it’s a condition for which we have treatment options. It’s a problem that can be solved. We hope to prospectively apply this in a real world environment. An artificial intelligence solution using chest imaging can play a big role in tackling TB.”
Gold Member
Ultrasound System
FUTUS LE
Gold Member
Electrode Solution and Skin Prep
Signaspray
Bladder Scanner
PBSV3.2
Forensic Imaging System
EXERO-DR
Latest Imaging IT News
- New Google Cloud Medical Imaging Suite Makes Imaging Healthcare Data More Accessible
- Global AI in Medical Diagnostics Market to Be Driven by Demand for Image Recognition in Radiology
- AI-Based Mammography Triage Software Helps Dramatically Improve Interpretation Process
- Artificial Intelligence (AI) Program Accurately Predicts Lung Cancer Risk from CT Images
- Image Management Platform Streamlines Treatment Plans
- AI-Based Technology for Ultrasound Image Analysis Receives FDA Approval
- AI Technology for Detecting Breast Cancer Receives CE Mark Approval
- Digital Pathology Software Improves Workflow Efficiency
- Patient-Centric Portal Facilitates Direct Imaging Access
- New Workstation Supports Customer-Driven Imaging Workflow
Channels
Radiography
view channel
Novel Breast Imaging System Proves As Effective As Mammography
Breast cancer remains the most frequently diagnosed cancer among women. It is projected that one in eight women will be diagnosed with breast cancer during her lifetime, and one in 42 women who turn 50... Read more
AI Assistance Improves Breast-Cancer Screening by Reducing False Positives
Radiologists typically detect one case of cancer for every 200 mammograms reviewed. However, these evaluations often result in false positives, leading to unnecessary patient recalls for additional testing,... Read more
MRI
view channel
PET/MRI Improves Diagnostic Accuracy for Prostate Cancer Patients
The Prostate Imaging Reporting and Data System (PI-RADS) is a five-point scale to assess potential prostate cancer in MR images. PI-RADS category 3 which offers an unclear suggestion of clinically significant... Read more
Next Generation MR-Guided Focused Ultrasound Ushers In Future of Incisionless Neurosurgery
Essential tremor, often called familial, idiopathic, or benign tremor, leads to uncontrollable shaking that significantly affects a person’s life. When traditional medications do not alleviate symptoms,... Read more
Two-Part MRI Scan Detects Prostate Cancer More Quickly without Compromising Diagnostic Quality
Prostate cancer ranks as the most prevalent cancer among men. Over the last decade, the introduction of MRI scans has significantly transformed the diagnosis process, marking the most substantial advancement... Read more
Ultrasound
view channel
Deep Learning Advances Super-Resolution Ultrasound Imaging
Ultrasound localization microscopy (ULM) is an advanced imaging technique that offers high-resolution visualization of microvascular structures. It employs microbubbles, FDA-approved contrast agents, injected... Read more
Novel Ultrasound-Launched Targeted Nanoparticle Eliminates Biofilm and Bacterial Infection
Biofilms, formed by bacteria aggregating into dense communities for protection against harsh environmental conditions, are a significant contributor to various infectious diseases. Biofilms frequently... Read more
Nuclear Medicine
view channel
New SPECT/CT Technique Could Change Imaging Practices and Increase Patient Access
The development of lead-212 (212Pb)-PSMA–based targeted alpha therapy (TAT) is garnering significant interest in treating patients with metastatic castration-resistant prostate cancer. The imaging of 212Pb,... Read more
New Radiotheranostic System Detects and Treats Ovarian Cancer Noninvasively
Ovarian cancer is the most lethal gynecological cancer, with less than a 30% five-year survival rate for those diagnosed in late stages. Despite surgery and platinum-based chemotherapy being the standard... Read more
AI System Automatically and Reliably Detects Cardiac Amyloidosis Using Scintigraphy Imaging
Cardiac amyloidosis, a condition characterized by the buildup of abnormal protein deposits (amyloids) in the heart muscle, severely affects heart function and can lead to heart failure or death without... Read more
General/Advanced Imaging
view channel
New AI Method Captures Uncertainty in Medical Images
In the field of biomedicine, segmentation is the process of annotating pixels from an important structure in medical images, such as organs or cells. Artificial Intelligence (AI) models are utilized to... Read more.jpg "Image: CT coronary angiography for diagnosing coronary artery disease offers advantages over other diagnostic tests (Photo courtesy of 123RF)")
CT Coronary Angiography Reduces Need for Invasive Tests to Diagnose Coronary Artery Disease
Coronary artery disease (CAD), one of the leading causes of death worldwide, involves the narrowing of coronary arteries due to atherosclerosis, resulting in insufficient blood flow to the heart muscle.... Read more
Novel Blood Test Could Reduce Need for PET Imaging of Patients with Alzheimer’s
Alzheimer's disease (AD), a condition marked by cognitive decline and the presence of beta-amyloid (Aβ) plaques and neurofibrillary tangles in the brain, poses diagnostic challenges. Amyloid positron emission... Read more.jpg "Image: Exemplary illustration of the labeling and segmentation process (Photo courtesy of TUM)")
CT-Based Deep Learning Algorithm Accurately Differentiates Benign From Malignant Vertebral Fractures
The rise in the aging population is expected to result in a corresponding increase in the prevalence of vertebral fractures which can cause back pain or neurologic compromise, leading to impaired function... Read moreIndustry News
view channel
Bayer and Google Partner on New AI Product for Radiologists
Medical imaging data comprises around 90% of all healthcare data, and it is a highly complex and rich clinical data modality and serves as a vital tool for diagnosing patients. Each year, billions of medical... Read more
