EIRL | NEXT MEDICAL VISION

News

Vision

Another set of powerful eyes to support future medical diagnosis.: EIRL is the next medical image diagnostic support technology to catalyze the coming generation of medical diagnosis. Together with the constant advancements in modalities, comes the challenges for doctors to interpret an exponentially large number of images for diagnosis. EIRL aims to provide an accurate, accelerated and streamlined solution for medical diagnosis by analyzing a variety of information required for diagnosis. EIRL will be the closest partner to all doctors, and serve as another powerful set of eyes to provide top quality medical care. To all doctors around the world, EIRL.

Technology

1. Expert-approved Training Data

EIRL is powered by knowledge derived from experienced doctors. All the data used to train EIRL is double and triple checked by experienced doctors. The selection process is further refined as only the images that led to an accurate diagnosis are used to train EIRL.

2. Learning from Limited Data

EIRL is instilled with LPixel's unique active learning technology, namely CARTA, and other powerful tools to enable EIRL to generate training data with great precision and efficiency, even if the number training data available on hand is limited.

3. Accommodation to Images of Varying Quality

In clinical practice, images for diagnostic purposes are taken with an array of different image acquisition tools and protocols. With aims to yield accurate results, EIRL is trained with a variety of training data acquired from a wide range of modalities and protocols.

4. Effortless Integration with PACS

Our goal is to integrate EIRL with the PACS that is currently being used in your institution.

Research & Development

Brain Aneurysm Detection
Intracranial Stenosis Detection
Quantitative Analysis of DESH
Cerebral White Matter Lesion Segmentation
Breast Cancer
Chest X-Ray
Lung Cancer
Liver Cancer
Colon Cancer
3D Construction of Pathology Images

*The listed themes are under development, and cannot be used as a medical device.

Brain Aneurysm Detection

With aims to keep the number of overlooked aneurysms to a minimum, our technology works on MRA scans to detect patterns that are similar to vascular deformations and unruptured brain aneurysms.
Intracranial Stenosis Detection

In efforts to reduce the number of misdiagnosed cases of intracranial stenosis, our technology conducts a careful analysis of MRA scans to detect patterns that are similar to narrowed arteries.
Quantitative Analysis of DESH

The Evans index and callosal angle are both very important markers for the diagnosis of Disproportionately Enlarged Subarachnoid Space Hydrocephalus (DESH) among patients with idiopathic normal pressure hydrocephalus (iNPH). Our goal is to develop the technologies capable of automatically calculating the Evans index and the callosal angle.
Cerebral White Matter Lesion Segmentation

This technology aims to analyze brain MRI scans to detect areas that are suspected to contain white matter lesions.
Breast Cancer

This technology conducts a time-intensity curve (TIC) analysis for MRI scans of the breasts. Our goal is to be able to augment machine learning technology to identify the degree of malignancy of the tumor.
Chest X-Ray

Trained with images reviewed by radiologists, this technology aims to streamline the chest X-ray screening process by instantly identifying the suspicious regions.
Lung Cancer

Currently under development with the National Cancer Center in Japan, this cutting edge technology targets lung cancer CT scans to automatically identify the type of lung cancer and gene mutation. Our goal is to reduce the need for highly invasive biopsies and open doors to imaging-based treatment options for patients.
Liver Cancer

This technology aims to analyze and monitor MRI scans of the liver over a period of time to look for regions that are suspected to contain tumors.
Colon Cancer

With aims to decrease the number of overlooked lesions, this technology analyzes the examination results of the colonoscopy images in real time and displays the areas in the colon where the prevalence of lesions, such as tumors, are high. This technology is being developed in collaboration with the Jikei University School of Medicine in Tokyo.

*The images used in the demo video was provided by Jikei University.
3D Construction of Pathology Images

Three-dimensional construction of pathological images is performed by interpolation processing and positioning, making it possible to track the three-dimensional structure of the cell. Our aim is to integrate deep learning to this three-dimensional information to create a powerful diagnostic tool that can help to decrease the number of overlooked lesions. This project is led by the Ministry of Economy, Trade and Industry in Japan as part of the Strategic Core Technology Advancement Program. LPixel is working in collaboration with TCK Co., Ltd. and others.

*The pathological samples used in the demo video was provided by Kurume University.