Related Experiment Video
Updated: Jan 17, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
Loss-Attention Physics-Informed Neural Networks for Enhanced Deepfake Detection in Medical Image Classification: A
Roshani A Parate1, Kirti Jain2
1Computer Science and Engineering, Sanjeev Agrawal Global Educational University, Bhopal, India.
Summary
A new method, MDD-MI-LAPINN, accurately detects medical deepfakes using Implicit Unscented Particle Filter (IUPF) and Loss-Attention Physics-Informed Neural Network (LAPINN). This approach ensures reliable medical images and prevents misdiagnoses.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Identifying deepfakes in medical imaging is critical for diagnostic accuracy and preventing fraud.
- High computational costs and unnoticed fraudulent changes pose significant challenges in medical deepfake detection.
Purpose of the Study:
- To introduce MDD-MI-LAPINN, a novel method for detecting deepfakes in medical images.
- To enhance the reliability and authenticity of medical image classification.
Main Methods:
- Medical images from Knee Osteoarthritis and Lung CT Scan datasets were pre-processed using the Implicit Unscented Particle Filter (IUPF).
- Classification was performed using a Loss-Attention Physics-Informed Neural Network (LAPINN) to differentiate authentic from manipulated images.
Main Results:
- MDD-MI-LAPINN achieved high accuracy (99.34% for real, 98.74% for fake images), precision, recall, and F1-scores.
- The method demonstrated superior performance compared to EMDD-MI-YOLO, MDD-MedNet, and DD-AGAA-CNN.
- Consistent and accurate detection and classification of both real and fake medical images were observed.
Conclusions:
- MDD-MI-LAPINN offers superior accuracy and robustness in medical deepfake detection.
- The integration of IUPF and LAPINN ensures precise and reliable medical image classification.