Gallbladder disease diagnosis from ultrasound using squeeze-and-excitation capsule network with convolutional bidirectional long short-term memory

2026-01-03 · Source: Machine learning : nature.com subject feeds · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Advanced, medium

Summary

A Hybrid Deep Learning Model with Feature Engineering for the Accurate Diagnosis of Gallbladder Disease Types (HDLMFE-ADGDT) has been proposed to improve the early identification and classification of gallbladder diseases from ultrasound images. This approach addresses the labor- and time-intensive nature of traditional ultrasound diagnosis. The HDLMFE-ADGDT technique first applies a Non-Local Means (NLM) filter for noise reduction and image quality enhancement. Subsequently, a Squeeze-and-Excitation Capsule Network (SE-CapsNet) is utilized for feature extraction. Finally, a hybrid Convolutional Neural Network with Bidirectional Long Short-Term Memory (CNN-BiLSTM) performs the actual diagnosis. Validated on a dedicated Gallbladder diseases dataset, the model achieved high performance metrics: 99.09% accuracy, 95.83% precision, 95.87% sensitivity, and 99.49% specificity, outperforming existing methodologies.

Key takeaway

For research scientists developing medical imaging diagnostics, this work demonstrates a highly effective hybrid deep learning architecture for gallbladder disease classification. You should consider integrating Non-Local Means filtering, Squeeze-and-Excitation Capsule Networks, and CNN-BiLSTM combinations into your models to achieve superior diagnostic performance on ultrasound data, especially where early and accurate identification is critical for patient outcomes.

Key insights

A hybrid deep learning model significantly improves gallbladder disease diagnosis from ultrasound images.

Principles

• Image pre-processing enhances diagnostic accuracy.
• Feature engineering is crucial for medical image classification.

Method

The HDLMFE-ADGDT method uses NLM filtering, SE-CapsNet for feature extraction, and CNN-BiLSTM for classification of gallbladder disease types from ultrasound images.

In practice

• Apply NLM filters for medical image denoising.
• Integrate SE-CapsNet for robust feature extraction.
• Combine CNN and BiLSTM for sequential image analysis.

Topics

• Gallbladder Disease Diagnosis
• Hybrid Deep Learning
• Squeeze-and-Excitation Capsule Network
• CNN-BiLSTM
• Ultrasound Image Analysis

Code references

• zgenz1537-code/GallBladderDiseaseDet

Best for: Computer Vision Engineer, Research Scientist, AI Researcher, Deep Learning Engineer, AI Scientist

Related on AIssential

• RARE disease detection from Capsule Endoscopic Videos based on Vision Transformers — Vision Transformers can be fine-tuned for multi-label classification of gastrointestinal diseases from capsule endoscopic videos.
• Hybrid deep learning model for multimodal vocal and lung signal analysis in health monitoring — A hybrid deep learning model effectively integrates vocal and lung signal analysis for accurate disease detection.
• Attention-Gated Convolutional Networks for Scanner-Agnostic Quality Assessment — A hybrid CNN-Attention model effectively assesses MRI quality, generalizing across diverse scanners without retraining.
• ECG Signal Classification Using Image-Based CNN Deep Learning — ECG signals can be effectively classified for cardiac abnormalities by transforming them into "2-D scalogram images" for CNN deep learning.
• Trusted Multi-View Deep Learning Classification of Fetal Congenital Heart Disease with Feature-level and Decision-level Fusion — The framework integrates multi-view echocardiographic data with uncertainty handling for robust fetal CHD classification.
• Cut instance mixing: A domain-specific data augmentation method applied to gastrointestinal lesion detection — A novel domain-specific data augmentation method, Cut Instance Mixing (CIM), has been introduced to address the scarcity of well-annotated medical image datasets and the subtle appearance of gastroint…

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.