FUJIFILM

The technology controls image quality based on a statistical model, an object model and a physical model using iterative processing.
This technology provides a high visibility image even at a low radiation dose , by maintaining the original texture of the image at a high denoise condition.
It brings both “high visibility at low dose” and “original texture at high noise reduction”

We are developing the AI image reconstruction technology that removes noise components of image by preventing the structure and the contrast from degrading by a high speed or a high resolution MR scanning. As a result, the technology shortens the scan time with keeping the image quality, or improves the visibility of the image by improving the image quality.

This flitering technology with AI suppresses aliasing noises which occur in extended images.

The technology virtually generates thin slices from thick slices. It can be applied to the whole body, useful for utilizing past data. It brings high visibility in the VR display and reconstructed sagittal/coronal images.

AI technology to segment and quantify the volume of each brain regions.
This technology can be used for pre-surgery simulation or calculation of the atrophy rate for each region between past and current exams.

This technology visualizes the bone density temporal difference by performing image registration between the past and current image of the same patient.
The increase and decrease in CT value will be highlighted.

Automatically extracts the pancreas, each surrounding organ area, and surrounding vessels from contrast-enhanced CT images and analyzes them in 3D.
The pancreatic duct diameter, residual pancreatic volume, measurement of the cross-sectional area, and resection simulation can be performed to support pancreatic surgery.

AI technology to extract each kidney and surrounding vessels from CT or MRI images assists volume calculation of renal and compare current and past results.
This technology can also precisely extract cystic kidneys, a genetic disease that is often difficult to extract due to the diversity of shapes, and assist diagnosis and treatment.

AI technology to automatically extract liver vessels and then labels liver according to the anatomic liver segments (Couinaud segments) based on the vessel running.
This can be used for Interventional Radiology planning for liver.

Technology that automatically extracts the liver parenchyma and surrounding vessels from CT images and analyzes them in 3D.
This technology can assist preoperative simulation by planning resection area based on the calculation of the dominant vessels.

AI technology to automatically extract lung fields, 5 lobes, and bronchus,　surrounding vessels from CT images.
From each extraction result, the percentage of low attenuation area is calculated.
These are expected to contribute the diagnosis of COPD.

This technology automatically extracts the femur, tibia, patella, cartilage, and meniscus from MRI images.
By measuring the thickness of the cartilage and the distance between the meniscus and tibia, it is possible to evaluate the condition of the knee joint without invasion.

Technology to recognize each vertebral body individually and label the vertebrae and ribs. This reduces the burden on the physician of visually counting bone positions. It is also used for visualization of fractures and bone metastasis by subtraction of rigid body alignment of vertebral bodies.

AI technology to subdivide lung into 10 segments in right lung, 8 segments in left lung from CT images. It can be used for confirming the position of chest nodules detected by a doctor.

AI technology to segment relatively large mediastinum, axillary lymph node from CT images. This technology can be used for assessment of lymph node enlargement.

AI technology to detect a suspcious rib fracture from CT images. This technology will assist prevention of overlooking of subtle rib fracture.

AI technology to identify suspicious region with COVID-19 related findings from CT images. This technology will help doctors diagnose efficiently.

AI technology to detect and quantify suspicious lesion from CT Images.
This will assit prevention of overlooking of nodule and generation of language of findings for radiology report.

This technology identifies various findings of interstitial pneumonia that appear on CT images, such as consolidation, reticular pattern, ground glass opacity and honeycomb, and calculates their distribution and volume.
This will assist in the diagnosis of the severity and therapeutic efficacy of interstitial pneumonia, which are conventionally performed qualitatively, by providing a quantitative value for assessment.

This technology extracts high-signal and low-signal areas in head CT images by comparing the left and right sides of the brain region. Generally, high-signal and low-signal areas are used to evaluate the state of hemorrhage and ischemia in the brain for stroke diagnosis. This will assist the diagnosis of head CT imaging.

The technology to extract and highlight areas of higher or lower absorption than the surrounding tissue. For example, high/low absorption in the liver, high absorption in the thoracic cavity, and low absorption in the kidney may be informative in determining the findings of each part. In the future, we aim to highlight both contrast-enhanced and non-contrast-enhanced images.

The technology detects three types of imaging findings: nodule, consolidation, and pneumothorax from chest X-ray images. It is expected to contribute to preventing oversights in various chest X-ray examinations, such as health checkups and routine medical examinations.

This technology extracts the target area from the image.
Positioning during imaging is performed automatically.
Unnecessary tissue is automatically removed before creating a 3D image for diagnosis.
It is a technology that supports the automation of inspections.

We developed AI algorithms to predict clinically important outcomes using inpatient data with pneumonia aggregated throught our integrated medical care support platform system. This technology will assist preparing personalized medical documents and optimizing medical resources

The technology supports populating radiology report of lung nodules based on the observation and analysis results of the properties. It enables to analyze multiple properties such as shape, absorption, presence of calcification, etc., and provide multiple finding sentence candidates using natural language processing technology.

The technology analyzes the properties of areas of interest in the liver and assists doctors in writing radiology report. It is possible to analyze not only the location and size, but also the appearance/disappearance of enhancement based on multi-phase images of contrast-enhanced CT. In addition, similar to lung nodule characterization analysis, it is also possible to present multiple finding sentence candidates using natural language processing technology.

The technology estimates high-value threshold of the region of interest, and quantify high intensity area in lung field. For example, it is expected to offer information for quantitative analysis of partially solid nodules.

This technology automatically measures EvansIndex, corpus callosum angle, and MidlineShift from head CT. EvansIndex and corpus callosum angle are expected to support the evaluation of hydrocephalus and MidlineShift to evaluate head trauma.

The technology under development quantitative detection of slight brain volume loss and iron deposition in the early stages of dementia using hybrid analysis of QSM（Quantitative Susceptibility Mapping）and VBM（Voxel Based Morphometry）in MRI. This will assist the diagnosis of dementia.

Technology that displays areas of high and low absorption in the liver, kidneys, and spleen compared to surrounding tissues.

The technology for segmenting the rectum and surrounding organs from MRI images, which is expected to be useful for surgical simulation in the lower digestive tract area.

The technology that utilizes AI to help detect findings suggestive of pancreatic cancer from abdominal CT images.

We have developed AI technology to identify areas suspected of cerebral hemorrhage or cerebral infarction in CT images of the head.
This technology is expected to aid in the diagnosis of stroke by helping to evaluate bleeding and ischemia in the brain.

We have developed AI technology to extract each region of the cerebrospinal fluid cavity from MRI images.
The aim is to improve diagnostic accuracy for Hakim's disease (iNPH), a form of dementia that is important to detect early for treatment.
This technology is expected to improve the objectivity and accuracy of diagnosis by enabling AI to efficiently analyze regions associated with key findings (DESH) and support differentiation from brain atrophy.

The technology that displays areas of low absorption in the pulmonary artery compared to surrounding tissues.
This is expected to aid in the diagnosis of pulmonary embolism.

The technology that automatically measures the height of each vertebral body and displays the results classified according to user-defined thresholds.
This technology is expected to aid in the diagnosis of vertebral fractures.

The technique for labeling vertebral numbers on MRI images.

When creating documents such as discharge summaries, we developed a technology that extracts necessary text from medical records to assist.

Based on clinical data accumulated in “CITA Clinical Finder,”
we have developed technology that predicts the risk of falls among outpatients, provides risk scores, and identifies factors contributing to AI predictions.

We have developed technology that predicts and lists medical data that is likely to be viewed based on patient information accumulated in “CITA Clinical Finder” and the viewing status of users, thereby supporting the reference of medical data.

The technology that highlights areas of high/low absorption in the adrenal glands, pancreas, gallbladder, and lymph nodes compared to surrounding tissues.

High signal/low signal region extraction technology, brain region labeling technology, brain extraction technology.

Observe the liver and surrounding organs while deforming them, enabling estimation of the position of blood vessels during surgical detachment.

Display 3D images created in the target application on a spatial reproduction display, and operate them in synchronization with our 3D system.

Displays the complex structure of the gallbladder and pancreas in three dimensions, supporting preoperative simulation of organ relationships and other factors.

Fujifilm has developed “Reading Report Structuring AI,” a unique natural language processing technology that structures reading reports created by radiologists for medical images captured by CT, MRI, and other imaging devices.
By structuring reading reports written based on the specialized knowledge of radiologists, it is possible to efficiently database and utilize reading reports that include medical jargon and expressions unique to physicians.

The technology utilizes a large-scale language model to automatically generate and present drafts of "summaries," which are documents that comprehensively describe patient information such as test results and treatment history.