US 12,437,390 B2
System and method for assessing quality of medical images
Mohamed Abdolell, Halifax (CA); Ning Hu, Ancaster (CA); Desmond Chung, Thornhill (CA); Ryan Duggan, Miramichi (CA); Dan Barzilai, Hod HaSharon (IL); Nicole Paquet, Middle Sackville (CA); and Amin Ghazanfari, Toronto (CA)
Assigned to Densitas Incorporated, Halifax (CA)
Appl. No. 17/295,408
Filed by Densitas Incorporated, Halifax (CA)
PCT Filed Nov. 25, 2019, PCT No. PCT/CA2019/051684
§ 371(c)(1), (2) Date May 19, 2021,
PCT Pub. No. WO2020/102914, PCT Pub. Date May 28, 2020.
Claims priority of provisional application 62/771,067, filed on Nov. 24, 2018.
Prior Publication US 2023/0071400 A1, Mar. 9, 2023
Int. Cl. G16H 30/40 (2018.01); G06T 7/00 (2017.01); G16H 30/20 (2018.01)
CPC G06T 7/0012 (2013.01) [G16H 30/20 (2018.01); G16H 30/40 (2018.01); G06T 2207/10088 (2013.01); G06T 2207/30068 (2013.01); G06T 2207/30168 (2013.01)] 26 Claims
OG exemplary drawing
 
1. A system of determining medical image quality to determine errors in obtaining medical images, the system comprising:
a memory unit, the memory unit storing: a) a plurality of predictive models that are deep learning models including Convolutional Neural Networks (CNNs) trained for different quality metrics, and b) a database for storing a plurality of predicted image quality scores;
a display device;
a processing unit in communication with the memory unit and the display device, the processor unit having a processor being configured to:
receive a first medical image and an associated plurality of image metadata;
determine a predicted image quality score based on the first medical image and one of the predictive models that receives at least two inputs based on parameters or parameter features, the predicted image quality score providing an indication of a probability of a particular non-conforming condition being present in the first medial image and/or extract a feature from the first medical image using the CNN to learn different quality metrics, wherein a predicted plurality of image quality parameter scores is determined with the parameters or parameter features;
map a given predicted image quality parameter score to a predicted image parameter index based on: (a) applying a threshold to the given image quality parameter score, where the threshold is based on an operating point on a receiver operator characteristic curve for the predictive model that was used to generate the given predicted image quality parameter score or (b) applying a user configurable predicted image quality parameter feature threshold;
generate a graphical user interface;
provide an output to the display device, the output including the graphical user interface, the indexed image quality parameter score, and the predicted image quality score;
store the predicted image quality score in the database; and
perform statistical process control on the predicted image quality scores stored on the database to isolate errors and perform root case analysis for increase in error rates.