Last Updated:
25 January 2025

TBXNet++ is an advanced AI-powered software for efficient tuberculosis (TB) detection in chest X-rays. It classifies cases as either normal (no TB) or abnormal (presence of TB) and provides a confidence score in percentage for each prediction. The software includes heatmaps to visually highlight affected lung areas, aiding clinicians in diagnosis. Designed for adaptability, TBXNet++ operates seamlessly in online and offline environments, making it suitable for various healthcare settingsy. With a streamlined focus on TB detection, it supports high-throughput screening and timely clinical decision-making. By integrating TBXNet++, healthcare providers can improve diagnostic accuracy, enhance workflow efficiency, and contribute effectively to TB control efforts. Demo and support services are available.
Certification
Pending
Development Stage
On the Market
Deployment
Online & Offline
Intended Age Group
16+ years
Target Setting
Primary health centres, general hospital (above primary level), teleradiology companies, government/public sector, e.g. national TB programme, private sector
Current Market
Pakistan
Input
Chest X-ray image format: JPG, PNG
Chest X-ray type: Posterior-anterior chest X-ray, Anterior-posterior chest X-ray
Output
Output includes:
Heatmap
Dichotomous output indicating whether TB is likely present or absent
Probability score for TB
Probability score for each abnormality
Location of each abnormality
Results are displayed in a tabular format, with an option for clients to export detailed PDF reports for convenient sharing and record-keeping.
Lung abnormalities included in the TB Score: Cavity, Consolidation, Opacity
Additional findings reported by the product:Cavity, Consolidation, Fibrosis, Nodule, Opacity, Prominence in hilar region


Hardware
The product is available online at out website and no hardware requirement for using.
For on-premise settings, TBXNET++ can either be deployed on a computer or laptop with minimum 8 GB memory. Please contact via email for more information. ilm@szabist-isb.edu.pk
Server
TBXNET++ can be deployed on cloud servers built by cloud hosting partners with the highest standards for privacy and data security. We also deploy on-premise servers that meet required specifications of 16GB memory or above and 8GB graphic card. Internet bandwidth of 5Mbps or above.

Hardware
The product is available online at out website and no hardware requirement for using.
For on-premise settings, TBXNET++ can either be deployed on a computer or laptop with minimum 8 GB memory. Please contact via email for more information. ilm@szabist-isb.edu.pk
Integration with X-ray Systems
Integration with PACS and Legacy Systems
It can be integrated with third party X-ray system. An API can take the x-ray system file from the software to our system for prediction.
Please contact via email for details. ilm@szabist-isb.edu.pk
Software
No special software is required incase of online usage,For Offline Usage Windows 10, minimum 8 GB RAM
Processing Time
Around 0.06 - 0.24 seconds per X-ray
Data Sharing & Privacy
TBXNet++ ensures complete data privacy —no patient information is stored. However, X-ray images are stored on our secure web server. X-ray images are never shared without consent of the patients.
Price
Pricing is flexible, with multiple subsciptions options available for customers. Both one time susbscription or per x-ray image scan option is available. Please contact via email or call for detailed information or quotation.
Software Updates
Every 6 months
Product Development Method
Deep Learning, CNN, Transformers
Training
The product was trained on 17,000 chest X-rays.
Reference Standard
Human reader (radiologists specializing in TB)
Publications
1. Iqbal, Ahmed, Muhammad Usman, and Zohair Ahmed. "Tuberculosis chest X-ray detection using CNN-based hybrid segmentation and classification approach." Biomedical Signal Processing and Control 84 (2023): 104667. https://doi.org/10.1016/j.bspc.2023.104667
2. Iqbal, Ahmed, Muhammad Usman, and Zohair Ahmed. "An efficient deep learning-based framework for tuberculosis detection using chest X-ray images." Tuberculosis 136 (2022): 102234.
https://doi.org/10.1016/j.tube.2022.102234
3.Asad, Muhammad, Azhar Mahmood, and Muhammad Usman. "A machine learning-based framework for Predicting Treatment Failure in tuberculosis: A case study of six countries." Tuberculosis 123 (2020): 101944.
This website works best with browsers other than Internet Explorer.