| CPC G06T 7/0014 (2013.01) [G06V 10/776 (2022.01); G06V 10/82 (2022.01); G16H 10/60 (2018.01); G06T 2207/10116 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30061 (2013.01); G06V 2201/03 (2022.01)] | 12 Claims |
|
6. A system for detecting pneumonia in chest X-ray images comprising:
a data acquisition module configured to obtain a dataset of medical images having a set of pneumonia-infected lung images and a set of healthy lung images;
a pre-processing module configured to standardize and partition said dataset of medical images into a training subset, a validation subset, and a testing subset;
a model training module comprising at least one convolutional neural network (CNN) configured for image processing and object detection, wherein said model training module is configured to fine-tune said CNN using said training subset to optimize performance of said CNN for pneumonia detection;
a model evaluation module configured to evaluate said model training module using said validation subset and then compute a set of performance metrics; and
a simulation module configured to execute a computation and a simulation on a cloud-based platform;
wherein said system is configured to detect a pneumonia condition from said chest X-ray image and said model training module utilizes an ensemble combination of at least two deep learning models;
wherein said model evaluation module computes said set of performance metrics comprising:
an accuracy determined from the percentage of a set of pneumonia predictions correct;
a precision determined from said set of pneumonia predictions by calculating the ratio of true positive pneumonia predictions to all positive pneumonia predictions;
a recall determined from an ability of said system to detect all actual pneumonia cases by computing the ratio of true positive pneumonia predictions to the total real pneumonia cases; and
an F1-score determined from said precision and said recall;
wherein said system is configured to provide early and accurate pneumonia screening, thus enabling life-saving treatment.
|