top of page

Product:

Inspectra CXR

Version 4

Company HQ: 

723 Supakarn Building, L2-L3,Chareonnakhorn Road, Klong Ton Sai, KlongSan, Bangkok 10600

Download Product Profile

Last Updated:

25th January 2025

Inspectra CXR is an AI-based software designed to analyze adult frontal chest X-rays for signs of eight abnormality conditions, including Tuberculosis. The software assists doctors in prioritizing and triaging chest X-ray images. In addition to offering top-tier accuracy, Inspectra CXR provides flexible installation solutions, full integration with PACS systems, or a standalone AI server, all meeting world-class security standards.

Currently, Inspectra CXR identifies features suggestive of the following eight abnormality conditions:
1. Tuberculosis
2. Pleural effusion
3. Lung opacity
4. Pulmonary edema
5. Mass
6. Nodule
7. Cardiomegaly
8. Atelectasis

The system delivers AI results as a secondary capture on DICOM for PACS integration or as JPEGs for the Inspectra application. Both formats display the lesion locations in a heatmap, a list of AI findings, and abnormality scores corresponding to each finding.

Certification

1. Thai FDA Notification Certificate (Approved)
2. HSA Singapore Authorization (Approved)

Development Stage

On the Market

Deployment

Online & Offline

Intended Age Group

Adult chest radiographs, starting from 15 years old.

Target Setting

Primary health centers
General hospitals (above primary level)
Teleradiology companies
Government/public sector, e.g., national TB programs
Private sector

Current Market

Thailand


Input

Can be used to read images from any make and model of chest X-ray machine
Chest X-ray image format: DICOM, JPG
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

  • Dichotomous output indicating whether each abnormality is likely present or absent

  • Probability score for TB

  • Probability score for each abnormality

  • Location of each abnormality

The default threshold probability score is set at the optimal value for each abnormality condition, with 44% for Tuberculosis. This threshold cannot be adjusted by users but can be customized upon request.


Results are given in a structured report that presents the probability of abnormal findings as percentages and consists of three subsections:

1. Overall Abnormality Score: The score for the case across eight possible findings.

2. Tuberculosis Abnormality Score: The probability of signs of Tuberculosis.

3. Abnormality Scores for Other Findings: The probabilities for the remaining seven findings.


Lung abnormalities included in the TB Score: Calcification, Cavity, Consolidation, Fibrosis, Interstitial markings, Opacity


Additional findings reported by the product: Atelectasis, Blunted costophrenic angle, Lung opacity (includes consolidation, fibrosis, and interstitial markings), Pulmonary edema, Cardiomegaly, Mass, Nodule, Pleural effusion

Hardware

We offer various installation options, including on-premises and software-as-a-service (SaaS), as well as integration options with PACS. For more details, please contact info@perceptra.tech. 

Server

Server requirements vary depending on the chosen solution. Please contact info@perceptra.tech for more details.

Integration with X-ray Systems

Integration with PACS and Legacy Systems

Inspectra CXR is compatible with any digital radiography X-ray machines. It can be integrated with X-ray systems from leading global manufacturers such as FujiFilm, Siemens, GE, Philips, Shimadzu, Canon, and Carestream.

Inspectra CXR can be integrated with PACS systems using the DICOM protocol.

Software

Please contact info@perceptra.tech for more details.

Processing Time

The average processing time is less than 60 seconds per image.

Data Sharing & Privacy

Inspectra CXR ensures patients’ privacy with our best-practice security standard including:
1. HIPAA, GDPR, and PDPA compliance
2. At-rest and in-transit encryption
3. User authentication and access authorization
4. Machine hardening with CIS benchmark

Software Updates

Customers using Inspectra CXR as a service receive instant updates when a new product version is released. Notifications regarding updates will be sent automatically.
On-premises customers can sign up for maintenance services and software updates. Please contact us for details.

Price

Inspectra CXR offers a volume-based annual subscription plan, which includes license costs and all maintenance services. For installation, there is a one-time upfront fee that depends on the solution selected by the client. Please contact info@perceptra.tech for more information.

Product Development Method

The most recent version of Inspectra CXR was developed using supervised deep learning, specifically convolutional neural networks (CNNs).

Training

1.5 million CXR images from a global population.

Reference Standard

Inspectra CXR uses certified radiologist reports as the ground truth/reference standards. For TB cases, Inspectra CXR incorporates certified radiologist reports, sputum tests, culture tests, and CT scans as ground truth/reference standards.

Publications

Chaisangmongkon, W., Chamveha, I., Promwiset, T., Saiviroonporn, P. and Tongdee,

T., 2021. External Validation of Deep Learning Algorithms for Cardiothoracic

Ratio Measurement. IEEE Access.

(https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9502122)

Saiviroonporn, P., Rodbangyang, K., Tongdee, T., Chaisangmongkon, W., Yodprom,

P., Siriapisith, T., Wonglaksanapimon, S. and Thiravit, P., 2021. Cardiothoracic

ratio measurement using artificial intelligence: observer and method validation

studies. BMC Medical Imaging, 21(1), pp.1-11.

(https://bmcmedimaging.biomedcentral.com/articles/10.1186/s12880-021-00625-0)

Chamveha, I., Tongdee, T., Saiviroonporn, P., Chaisangmongkon, W. (2020). Local

Adaptation Improves Accuracy of Deep Learning Model for Automated X-Ray

Thoracic Disease Detection : A Thai Study. arXiv preprint arXiv:2004.10975.

(https://arxiv.org/pdf/2004.10975.pdf)

Chamveha, I., Promwiset, T., Tongdee, T., Saiviroonporn, P., Chaisangmongkon, W.

(2020). Automated Cardiothoracic Ratio Calculation and Cardiomegaly

Detection using Deep Learning Approach. arXiv preprint arXiv:2002.07468.

(https://arxiv.org/abs/2002.07468)

Does this page require updates? Send us a message:

This website works best with browsers other than Internet Explorer.

© 2023 Stop TB Partnership and FIND

bottom of page