MyAdvantech Registration

MyAdvantech is a personalized portal for Advantech customers. By becoming an Advantech member, you can receive latest product news, webinar invitations and special eStore offers.

Sign up today to get 24/7 quick access to your account information.

Advantech’s WISE-PaaS/AIFS AI Medical Imaging Solutions to Improve Diagnosis, Treatment, and Decision-Making

Medical imaging is an indispensable technology used in hospitals and radiologists play an invaluable role in in diagnosing and treating patients. Indeed, effective diagnosis relies on radiologists’ professional interpretation of X-rays, CT, or MRI images and scans. This is not an easy task, as the average proportion of images showing acute conditions is only 4 ~ 10%. This means, radiologists need to review every individual image carefully—resulting in long working hours, heavy workloads, and eye fatigue. Additionally, a first-come-first-served workflow result in patients with potentially life-threatening conditions not being diagnosed and treated in time.

Advantech developed its WISE-PaaS/AIFS (AI Framework Service) medical imaging solution and consulting services to meet the needs of medical centers. WISE-PaaS/AIFS is a one-stop industrial app platform for AI management, development, model training and deployment. One successful project was for a "Chest X-ray solution for Pneumothorax Detection" aimed at prioritizing and categorizing images using an AI assisted detection process. If the system detected a potentially dangerous lesion, it automatically sends a warning flag to the radiologist in charge. This helps medical professionals reprioritize their workloads, and enables them to concentrate on high-risk cases first—ensuring priority treatment for critically ill patients. Similarly, the solution helps radiologists by improving their working environments, compensating for any eye fatigue issues and prolonging their occupational working life.

Four Procedures for AI Model Optimization and Prognosis Improvement

Advantech’s WISE-PaaS/AIFS-based AI medical imaging solution comprises six implementation steps: image acquisition, reasoning, confirmation, labeling, model retraining, and model replacement. Continuously repeating these six steps refines and optimizes the AI’s learning model making it better at analysis and detection.WISE-PaaS/AIFS's AI medical imaging solution also includes four operating procedures in its model training process: 

First, image data pre-processing converts the image format from DICOM to PNG so radiologists can use AI labeling aids (AIAA) to assist, accelerate, and optimize labeling.

Second, image segmentation and modeling, is an AI modeling process based on detection requirements set by a doctor. It combines with Generative Adversarial Network (GAN) technology to simulate and synthesize training data to address issues endemic to hospitals—including the lack of image data and AI model training difficulties.

Third, feature extraction and analysis uses a collection of statistical methods to analyze the intensity, texture, and features of individual lesions. 

Fourth, diagnostic analysis panel helps doctors analyze lesions. This combination of procedures augments care by giving doctors diagnosis and treatment suggestions aimed at improving prognosis.

AI Medical Imaging Solution: Pneumothorax Detection and More in the Future

A large hospital in southern Taiwan is using Advantech’s WISE-PaaS/AIFS AI pneumothorax detection solution. The solution’s proven analytical capabilities optimize radiologists’ workloads by presenting pneumothorax lesion images first. Similarly, pneumothorax patients who have X-ray images taken at a later treatment stages benefit from priority diagnosis and treatment. This increases their recovery and survival rates. 

The AI deep learning pneumothorax X-ray imaging cluster, based on open data, was combined with data pre-processing and hyper-parameter AI framework services during deployment. These services optimized the X-ray image segmentation AI model to enable reliable pneumothorax diagnosis. A Sørensen-Dice Index was applied to the hospital’s test data set following implementation. It exceeded 0.8 — setting a benchmark for the official launch of the solution.

What’s more, the six implementation steps of the AI medical imaging solution— image acquisition, reasoning, confirmation, labeling, model retraining, and model replacement—are equally applicable to the detection of aortic dissection, pneumonia, tuberculosis, lung cancer, cerebral hemorrhage, and other diseases. Accordingly, Advantech will cooperate with more smart health ecosystem partners to expand the scope of its WISE-PaaS/AIFS medical imaging applications. 

WISE-PaaS & WISE-Marketplace

The WISE-PaaS data application platform end-to-cloud architecture provides a comprehensive development environment for data collection, analysis, management, and applications, while promoting interoperability between I.Apps via the WISE-Marketplace and accelerating the development of industrial solutions.


As an one-stop AI management platform, WISE-PaaS/AIFS supports diversified AI framework container, model retraining and redeployment, and edge-cloud collaborative management; which accelerates large-scale AI industrial applications.