US 12,224,047 B2
Systems and methods of radiology report processing and display
Zeljko Velichkovich, Carver, MN (US); Theodore J. Walker, Kansas City, MO (US); Patrick T McGaughey, Chanhassen, MN (US); Shwan Kim, Thousand Oaks, CA (US); Kimberley S. Miller, Mount Vernon, WA (US); and Timothy Braatz, Waverly, PA (US)
Assigned to Virtual Radiologic Corporation, Minneapolis, MN (US)
Filed by Virtual Radiologic Corporation, Minneapolis, MN (US)
Filed on Aug. 31, 2022, as Appl. No. 17/900,736.
Prior Publication US 2024/0071586 A1, Feb. 29, 2024
Int. Cl. G16H 15/00 (2018.01); G06F 40/20 (2020.01); G16H 50/70 (2018.01)
CPC G16H 15/00 (2018.01) [G06F 40/20 (2020.01); G16H 50/70 (2018.01)] 25 Claims
OG exemplary drawing
 
1. A method of processing prior findings data in a radiology review workflow, performed with electronic operations executed with a processor of a computing device, the electronic operations comprising:
receiving unstructured text from a prior radiology report of a patient, the prior radiology report prepared for the patient in a corresponding prior radiology study;
processing the unstructured text with a natural language processing (NLP) engine to classify prior findings in the prior radiology study, each of the prior findings indicating at least one observed condition and a characteristic of the at least one observed condition,
wherein the NLP engine includes a neural network that is pre-trained to classify the prior findings, wherein the neural network is pre-trained by associating patterns and words from unstructured text used in training radiology reports with training pathology phrases, and
wherein the neural network operates to classify the prior findings in the prior radiology study of the patient by:
parsing the unstructured text from the prior radiology report to identify pathology phrases that describe a pathology or existence of a respective medical condition; and
determining the at least one observed condition and the characteristic of the at least one observed condition from textual usage of the identified pathology phrases; and
presenting the prior findings in a radiology review user interface, the radiology review user interface to display the at least one observed condition and the characteristic of the at least one observed condition that is indicated by the prior radiology report.