We use cookies to understand how you use our site and to improve your experience. This includes personalizing content and advertising. To learn more, click here. By continuing to use our site, you accept our use of cookies. Cookie Policy.

Features Partner Sites Information LinkXpress hp
Sign In
Advertise with Us
GLOBETECH PUBLISHING LLC

Download Mobile App




Deep Learning Model Diagnoses Knee Abnormalities Like Experienced Radiologist

By MedImaging International staff writers
Posted on 26 Nov 2024
Print article
Image: The deep learning model can assist with classifying 12 common types of knee abnormalities (Photo courtesy of 123RF)
Image: The deep learning model can assist with classifying 12 common types of knee abnormalities (Photo courtesy of 123RF)

The knee joint, a complex hinge joint, is one of the primary load-bearing joints in the human body, facilitating various movements required for daily activities. Aging and injury can cause several knee abnormalities, leading to pain and dysfunction. Therefore, accurate diagnosis of these abnormalities is essential for creating tailored treatment plans and enhancing patients' quality of life. Given the knee joint's intricate anatomy, different scanning parameters can yield varied results. Additionally, some subtle lesions may be overlooked, especially by radiologists with limited experience. Multi-sequence knee magnetic resonance imaging (MRI) is an advanced, non-invasive technique for diagnosing knee pathologies. However, interpreting MRI results is time-consuming and heavily dependent on the expertise of the radiologist. In response to this, researchers have developed an innovative deep learning model that assists in classifying 12 common types of knee abnormalities, improving both efficiency and diagnostic accuracy.

In a collaborative study, researchers at HKUST Smart Lab (Kowloon, Hong Kong) and the Third Affiliated Hospital of Southern Medical University (Guangzhou, China) analyzed data from 1,748 patients. The study included T1-weighted (T1W), T2-weighted (T2W), and proton density-weighted (PDW) MRI sequences from sagittal, coronal, and axial planes. By combining these MRI data with arthroscopy results, which is considered the gold standard for diagnosing knee abnormalities, the researchers conducted a comprehensive analysis and identified 12 common knee abnormalities. They then developed a deep learning model, incorporating Co-Plane Attention across MRI Sequences (CoPAS), to classify these abnormalities. The model effectively captured intensity variations across different MRI sequences and identified complex correlations with abnormality types by separating spatial features, which resulted in high classification accuracy.

To assess the model's effectiveness, the researchers conducted simulated clinical testing. Radiologists first provided diagnoses based solely on MRI scans. After a washout period, the radiologists were asked to make diagnoses again, this time using the model’s output as a reference. The results showed that the model's average diagnostic accuracy exceeded that of junior radiologists and was comparable to senior radiologists. The overall diagnostic accuracy of all radiologists improved significantly with the model's assistance, according to the study published in Nature Communications. An additional interpretability analysis compared the clinical findings with the model’s output, revealing that the model’s decision-making process closely matched clinical preferences. This suggests that the model has developed a decision-making framework similar to that of radiologists, which enhances its reliability for clinical implementation.

“This innovative CoPAS model demonstrates diagnostic performance comparable to that of radiologists. It is particularly beneficial in bridging the gap between less experienced and senior doctors,” said Assistant Professor CHEN Hao from the Department of Computer Science and Engineering and Department of Chemical and Biological Engineering at HKUST, who led the study. “Our findings underscore the promise of artificial intelligence in healthcare, highlighting its potential to identify and validate new clinical insights.”

New
Gold Member
X-Ray QA Meter
T3 AD Pro
New
Portable Color Doppler Ultrasound System
S5000
New
X-ray Diagnostic System
FDX Visionary-A
New
Ultrasound Scanner
TBP-5533

Print article

Channels

Nuclear Medicine

view channel
Image: Example of AI analysis of PET/CT images (Photo courtesy of Academic Radiology; DOI: 10.1016/j.acra.2024.08.043)

AI Analysis of PET/CT Images Predicts Side Effects of Immunotherapy in Lung Cancer

Immunotherapy has significantly advanced the treatment of primary lung cancer, but it can sometimes lead to a severe side effect known as interstitial lung disease. This condition is characterized by lung... Read more

General/Advanced Imaging

view channel
Image: AI-enabled analysis of images meant to catch one disease can reveal others (Photo courtesy of 123RF)

AI Tool Offers Opportunistic Screening for Heart Disease Using Repurposed CT Scans

In the field of medical imaging, the term "opportunistic screening" refers to the repurposing of existing medical images by radiologists to diagnose illnesses beyond what the scan was originally meant to find.... Read more

Imaging IT

view channel
Image: The new Medical Imaging Suite makes healthcare imaging data more accessible, interoperable and useful (Photo courtesy of Google Cloud)

New Google Cloud Medical Imaging Suite Makes Imaging Healthcare Data More Accessible

Medical imaging is a critical tool used to diagnose patients, and there are billions of medical images scanned globally each year. Imaging data accounts for about 90% of all healthcare data1 and, until... Read more
Copyright © 2000-2024 Globetech Media. All rights reserved.