Related Experiment Video
Updated: Jul 4, 2026
08:36
Ultrasound Imaging-guided Intracardiac Injection to Develop a Mouse Model of Breast Cancer Brain Metastases Followed by Longitudinal MRI
Published on: March 6, 2014
18.9K
Advancing breast ultrasound diagnostics through hybrid deep learning models
Ajmeera Kiran1, Janjhyam Venkata Naga Ramesh2, Irfan Sadiq Rahat3
1Department of Computer Science and Engineering,MLR Institute of Technology, Dundigal, Hyderabad, Telangana, 500043, India.
Computers in Biology and Medicine
|August 14, 2024
Summary
The novel EfficientKNN model combines EfficientNetB3 and k-Nearest Neighbors for superior medical image classification, achieving 100% accuracy. This hybrid approach enhances early diagnosis and treatment planning.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Medical imaging is crucial for identifying abnormalities, enabling timely diagnosis and treatment.
- Accurate classification of medical images is essential for informed clinical decisions.
Purpose of the Study:
- To introduce the EfficientKNN model, a hybrid deep learning approach for enhanced medical image classification.
- To evaluate the performance of EfficientKNN against traditional models and EfficientNetB3 alone.
Main Methods:
- Utilized EfficientNetB3 as a feature extractor, followed by GlobalAveragePooling2D and optional Principal Component Analysis (PCA).
- Employed an optimized k-Nearest Neighbors (k-NN) algorithm for classification, fine-tuned via cross-validation.
- Trained the model on a curated dataset of medical images using data augmentation, segmentation, and ensemble optimization algorithms.
Main Results:
- EfficientKNN achieved 100% accuracy in multiple tests, outperforming VGG16, AlexNet, VGG19, and EfficientNetB3 alone in accuracy, precision, and F1-score.
- Demonstrated superior performance in key classification metrics.
Conclusions:
- The EfficientKNN model represents a significant advancement in medical image classification.
- The model shows potential for real-world diagnostic applications due to its accuracy, scalability, and efficiency.
- EfficientKNN offers improved diagnostic accuracy and clinical applicability by integrating deep learning with interpretable k-NN.
