Health Care

CASE Study 1

Athlete movement prediction for training

With Genexia’s improved decision-making, we were able to accurately predict the movement direction of athletes in 97% of the instances when the athlete is 1m away from the NPC defender. This case study was done in collaboration with Cincinnati Children’s Hospital.

There are many cases of concussions in athletes caused by collision. The solution was to rewire brain plasticity for improved decision-making to improve collision avoidance.

0 %

In the VR sport simulation, the goal was to accurately predict athlete decision making up to t = -3000 ms of an athlete/virtual opponent interaction.

FuzzyBolt was challenged against a well-validated, physics-based behavioral model of athlete navigation. At t = -1200 ms, FuzzyBolt (86% accuracy) began to outperform the physics-based model (75%) and continued to outperform the physics-based model by a consistent margin through t = 0. FuzzyBolt was able to attain an accuracy as high as 97% against the physics-based model with an accuracy of 83%.

CASE Study 2

Dermatology disease identification

Objective of this case study is to determine the type of Erythematous-Squamous Disease. The differential diagnosis of erythematous-squamous diseases is a real problem in dermatology. They all share the clinical features of erythema and scaling, with very little differences.

Usually a biopsy is necessary for the diagnosis but unfortunately these diseases share many histopathological features as well. Patients were first evaluated clinically with 12 features. Afterwards, skin samples were taken for the evaluation of 22 histopathological features. The values of the histopathological features are determined by an analysis of the samples under a microscope.

0 %

CASE Study 3

Breast Cancer Detection

Objective of this case study is to aid medical practitioners in the classification of breast tumors into malignant or benign, using the features extracted from digitized image of a fine needle aspirate.

Real valued features like radius, texture, compactness, concavity, symmetry, area etc. are computed for each cell nucleus. Dataset consists of 212 malignant and 357 benign cases

0 %

CASE Study 4

Mass and Calcification detection in Mammograms

Currently, work is progressing towards this goal. We have been able to get good results in highlighting breast masses and calcifications. Our image processing AI has been able to detect and segment mass and calcifications from breast mammography scans. As the presence of mass and calcification could be indicators for malignancy, it is extremely important for these to be detected as early as possible. Our AI was able to detect benign microcalcifications as well as malignant clustered microcalcifications and other types of calcifications. Since certain types of breast mass can be indication of breast cancer, early detection is important. Mass detection was also possible with our AI for digitized mammography scans. Further work is being conducted to enhance the predictions and make it universally applicable to different types of mammogram images that are used by radiologists today.

Mammogram based breast density estimation

We are also expanding the capabilities of our Image Processing AI to make accurate classification of the breast density of patients from mammogram images.  Radiologists find it difficult to evaluate patients who have high breast density. So, the ability to identify patients with high breast density helps to transfer such cases for further testing and allows radiologists to focus on low and medium dense mammograms that can be evaluated.

Calcifications detection

Left image shows the actual specks of calcifications present on a section of mammogram. Right image shows the precise detections made by our Image Processing AI

Mass detection

Left image shows the actual mass present on  mammogram. Right image shows the accurate detections made by our Image Processing AI